Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
AvionicsThe main contributor to innovation in aviation
Nuna5What went wrong?
Controlling the brainGraduation report
axwell
Edition 13.3Alumni edition
April 2010
M Magazine of the Electrotechnische Vereeniging
2 Maxwell 13.2 January 2010
HIER INVOEGEN:PAGINA2NEN.PDF
Maxwell 13.3 April 2010 3
From the BoardYou find in your possession this year’s alumni edition of Maxwell. This is an extra
thick edition of our quarterly magazine, which we are quite proud of. It represents an
important method for us to stay in contact with graduated electrical engineers. This
contact is important to us, as we aim to give people the opportunity to broaden their
view on their field of study and provide information about jobs and companies that
could be interesting to them. Henceforth, if you know an electrical engineer with an
interesting position, or if you would like to share your own experiences, please don’t
hesitate to contact us.
To follow, this quarter we made sure that aspiring electrical engineers got a chance
to open their minds to a wide variety of new ideas and projects. We went to CeBit in
Hannover, went on an excursion to Shell in Moerdijk, where we visited their chemi-
cal plant, and received a lunch lecture from Technolution. But of course, visiting
companies is not the only way to broaden your view on our profession. One big thing
we are working on at the moment is the organisation of the EESTEC exchange. The
European organisation EESTEC has the aim to bring electrical engineering students
from all over Europe together to give them the opportunity to experience foreign cul-
tures. In June there will be an exchange to Delft, which a committee of the ETV will
organise. To coordinate the activities of the EESTEC organization all members meet
once a year at a congress to speak about the plans for the organization for next year,
to elect a new EESTEC board and to sort out the financial affairs of the organization.
This year this congress took place in Athens where Delft was represented by two
The Board with the delicious cream pie
The Board visiting honoree member De Kroes
Jan and Frank in Athens
Board members of the ETV, Frank Teunisse and Jan Christiaanse. They had a great
week and gave a lot of useful input on the meetings.
That same week it was the ETV’s Dies. On the 23rd of March we had a huge home-
made pie in the faculty hall. It took two full days to make, but it was worth all the trouble. The pie was absolutely delicious.
Aside from all the activities the Board also visits the honoree members of the association. The ETV has 12 honoree members. These
people are usually visited once a year by the Board of the ETV. We have heard a lot of stories about their carreers and experiences,
what it was like to be a student and how they have been keeping themselves busy.
But most of the time you can find us in the Board room of the ETV. So if you like to meet some other students in a relaxed atmos-
phere, just drop along for a free cup of coffee.
On behalf of the inviting Board,
Imke Zimmerling, President
4 Maxwell 13.3 April 2010
ETV MAGAZINE “MAXWELL” Year 13 – edition 3 – April 2010 PRINTING Thieme Media Services, Delft NUMBER OF COPIES 5500 EDITORS Ben Allen, Joost van Driel, Benjamin Gardiner, Stephan Groot, Maarten Kastelein, Jeroen Ouweneel, Imke Zimmerling CONTACT Maxwell, p/a Electrotechnische Vereeniging, Mekelweg 4, 2628 CD Delft, phone: 015-2786189 or 015-2781989, fax: 015-27 81002, e-mail: [email protected], website: www.etv.tudelft.nl CHANGE OF ADDRESS Please send your changes to the address above, or use the website. Alumni can change their address via the Alumni Office website: www.alumni.tudelft.nl. ADVERTISEMENTS ASML (p.47), Farnell (p.48), Mastervolt (p.29), Navigon (p.35), NEN (p.2), Pinewood (p.35), Smit (p.15), Technolution (p.45), TNO (p.21) SUBSCRIPTIONS Non-members can receive the Maxwell four times a year, against a contribution of €10,- per year. For more information, please contact the Maxwell Committee.
Editorial
Although every single edition of the Maxwell on its own is special, once a year the entire Maxwell Committe prepares itself for her yearly magnum opus. By reading this, you’ve reached the fourth page of the Alumni edition of already the thir-teenth year of the Maxwell. For the committee the alumni edition always provides an extra chal-lenge, as it contains more pages to fill with inter-esting content than a regular edition.
What to expect of this Maxwell? Besides the usu-al sections like the From the Board, the news-flash and an overview of the recent activities of the Electrotechnische Vereeniging, the Maxwell Committee tried to provide you with an article about every branch of Electrical Engineering. From the Telecommunications angle an article about the communication network of the Dutch emergency services C2000 is presented. Marijn van Dongen provides a report about his master thesis work, by which also Microelectronics is represented. An insight in the development in the powering of trams is given by Professor van der Sluis from Power Engineering. Last, some Computer Engineering topics are touched upon by two different companies. The Austrian com-pany TTTech explains its adaptation of the exisit-ing Ethernet. The Dutch Technolution explained how difficult it is to use the GPU for other pro-cessing tasks.
To conclude, I expect that you will have some pleasant hours reading this edition of the Max-well. For any remarks, complaints or other co-ments, please feel free to contact the Maxwell Committee.
Stephan Groot
Contents
Controlling the brain
Avionics
12
22
Maxwell 13.3 April 2010 5
From the board 3 Newsfl ash 6 Educational announcements 8 FSR 8 Activities of the ETV 9 Cooking with: Rob Remis 28 TTEthernet 30
Joost may know it 34 Circuit 38 Per elektrische scooter naar Kopenhagen 40 Gadgets 42 Technolution: GPU 43 Column: Prof. Ligthart 46
C2000: Life-saving communication?
Innovatie intrammetjesland
NuNa 5:What went wrong?
16
26
36
6 Maxwell 13.3 April 2010
Newsfl ash
Skinput: a touchscreen on your armPaul Marks, New Scientist
Finding the keypad on your cellphone or music player a bit
cramped? Maybe your forearm could be more accommodating.
It could become part of a skin-based interface that effectively
turns your body into a touchscreen. Called Skinput, the sys-
tem is a marriage of two technologies: the ability to detect the
ultralow-frequency sound produced by tapping the skin with a
fi nger, and the microchip-sized “pico” projectors now found in
some cellphones.
The system beams a keyboard or menu onto the user’s forearm
and hand from a projector housed in an armband. An acous-
tic detector, also in the armband, then cal-
culates which part of the display you want
to activate. But how does the system know
which icon, button or fi nger you tapped?
Chris Harrison at Carnegie Mellon Univer-
sity in Pittsburgh, Pennsylvania, working
with Dan Morris and Desney Tan at Micro-
soft’s research lab in Redmond, Washington,
exploit the way our skin, musculature and
skeleton combine to make distinctive sounds when we tap on
different parts of the arm, palm, fi ngers and thumb.
They have identifi ed various locations on the forearm and hand
that produce characteristic acoustic patterns when tapped. The
acoustic detector in the armband contains fi ve piezoelectric
cantilevers, each weighted to respond to certain bands of sound
frequencies. Different combinations of the sensors are activated
to differing degrees depending on where the arm is tapped.
Twenty volunteers tested the system and most found it easy
to navigate through icons on the forearm and tap fi ngers to
actuate commands. Skinput works very well for a series of ges-
tures, even when the body is in motion, the
researchers say, with subjects able to scroll
through menus whether they moved up and
down or fl icked across their arm. The sys-
tem could use wireless technology like Blue-
tooth to transmit commands to many types
of devices. The researchers will present their
work in April at the ACM Computer-Human
Interaction meeting in Atlanta, Georgia.
Figure 1: The Skinput-system projecting
buttons on an arm.
MIT makes display with fl ying pixelsImagine that pixels could fl y out of your computer screen and
create an immersive, luminous cloud capable of displaying
digital information in three-dimensional space. This is the
vision beyond Flyfi re, a new project put together by research-
ers at MIT’s SENSEable City Lab and Aerospace Robotics and
Embedded Systems Laboratory (ARES Lab). Flyfi re uses a large
number of remotely controlled, self-organizing “micro helicop-
ters”. Each helicopter contains small LEDs and acts as a smart
pixel. Through digitally controlled movements, the helicopters
perform elaborate and synchronized choreographies, generating
a unique free-form display in three-dimensional space.
Using the self-stabilizing and precise controlling technology de-
veloped by the ARES Lab, the motion of the pixels is adaptable
in real time. The Flyfi re canvas can transform itself from one
shape to another or bring a two-dimensional photographic im-
age into an articulated shape (background image). “Today we
are able to simultaneously control a handful of micro helicop-
ters, but with Flyfi re we are aiming to scale up and reach very
large numbers,” said Emilio Frazzoli, head of the ARES Lab.
“Flyfi re opens up exciting possibilities: as on a conventional
screen, pixels can change color, but now they can also move,
creating a transient trace of light in three-dimensional space,”
said team member Carnaven Chiu. “Unlike traditional displays
that can only be seen from the front, Flyfi re becomes a three
dimensional immersive display that can be experienced from
all directions.” Flyfi re is conceived as a public space installa-
tion, in which the pixels recharge every few minutes and then
perform in space. “In general, there are two ways to increase
the resolution of a display,” said Carlo Ratti, director of the
SENSEable City Lab.
“One is to use smaller pixels.The other one is to look at it from
farther away. Flyfi re adopts the second approach to create a
unique visual experience in large public spaces.” Flyfi re is made
possible by recent advances in battery technology and wireless
control. It aims to be a step towards ‘smart dust’, the idea that
computing is becoming increasingly smaller, addressable, per-
vasive - and persuasive. The Flyfi re project was developed by
E. Roon Kang, Carnaven Chiu, Caitlin Zacharias, Shaocong
Zhou, Assaf Biderman and Carlo Ratti of SENSEable City
Lab in collaboration with Erich Mueller and Emilio Frazzoli of
ARES Lab.
Source: http://senseable.mit.edu/flyfire
Maxwell 13.3 April 2010 7
Record communication speeds over ceiling lightsResearchers at Fraunhofer Heinrich Hertz Institute in Berlin
together with their Siemens colleagues have scored a peak data
rate of 500 megabits per second (Mbit/s) using off-the-shelf
LED lights. The new benchmark breaks the previous record
they held of 200 Mbit/s. Data transport over visible light is a
means of transmission that is license-free, and tap-proof and
that opens the way for a range of novel applications in the
home, industry and transport.
Researchers at Siemens Corporate Technology in Munich and
the Heinrich Hertz Institute set the new
free space data transmission record for a
distance of up to 5 meters using a white
light emitting diode from the Siemens
subsidiary Osram. Data were directly
modulated from the supply current onto
the quantity of light emitted by the LED.
The Ostar LED used is one of the bright-
est now on the market and can be modu-
lated so rapidly that a highspeed
data transmission rate of 500 Mbit/s
can be achieved while the human
eye detects no change in the level
of brightness. The receiver is a photodetector that transforms
light signals into electrical impulses.
Visible Light Communication (VLC) is a medium that holds
out the promise of a wide range of applications. In the home
it can be a valuable extension to established WLAN technolo-
gies as in many buildings wireless networks are increasingly
impeded by clogging of the three independent WLAN frequen-
cy bands which leads to collisions between data packets. As a
license-free and previously unexploited medium, visible light
offers a viable alternative. A further key advantage is that VLC
also offers tap-proof secure lines as only the receiver located
directly in the light cone can receive the transmitted data, thus
precluding any “eavesdropping” on the light beam. In factories
and medical technology there is a need for data transmission in
places where wireless cannot be deployed or only to a limited
extent. Yet another area of application is in
the transport domain where LED traf-
fic lights and railway signals can relay
information to cars and trains.
The researchers also demonstrated
that a network of up to five LEDs is
capable of achieving data transmis-
sion speeds of up to 100 Mbit/s over a
longer distance. This is a critical point for prac-
tical applications as, for instance, data from ceiling lights
can then be sent to a receiver on a desktop no matter where
the desk is positioned in the room. Since 2007 the Institute of
Electrical and Electronics Engineers (IEEE) has been working on
standardization of the technology in a procedure scheduled for
completion by late 2010.
Source: http://www.hhi.fraunhofer.de.
Figure 2: The principle of
LED-light communication.
Babbage nanomachine promises low-energy com-putingNot only did Charles Babbage lay the foundations for the com-
puter revolution, his designs for mechanical computers also
provide a blueprint for energy efficiency. So say Raj Mohanty
of Boston University and his colleagues, who have created a
nanoscale mechanical logic gate that could form the basis of
tiny mechanical computers.
The gate consists of a strip of silicon 300 nanometres wide sit-
ting between two chunks of silicon. Applying a voltage between
one chunk and the strip causes the strip to vibrate, like the reed
in a clarinet. With the right voltage, the strip will enter a so-
called hysteretic regime - where it will vibrate with one of two
amplitudes.
“The two amplitude states are separated by a potential barrier,”
says Mohanty. By using a pair of electrical pulses that work
with the resonating strip to provide that kick in potential, the
team were able to flip the vibration from one amplitude to the
other. If just one of the pulses - or neither of them - resonates
with the strip then it remains in its existing vibrational state.
In other words, says Mohanty, the device acts as an AND logic
gate (Nano Letters, DOI: 10.1021/nl9034175).
While the gate is not as fast as its traditional equivalent, it loses
far less energy per operation, Mohanty claims. “When you have
millions of devices on a chip, energy loss adds up,” he says.
Trading speed for energy might be beneficial in some situations.
Source: http://www.newscientist.com/
Figure 3: The new chip, with inputs and output.
8 Maxwell 13.3 April 2010
Educational Announcements
MIGRATION TO OSIRIS
Osiris is now almost fully functional, and the migration of the
ISPs is also almost fi nished. Starting next quarter, subscrip-
tions for exams have to be done in Osiris. Furthermore, from
now on there will be a single deadline for subscribing for all
exams. For the fourth quarter, this deadline will be June 6th.
NEW TIMETABLE SOFTWARE IN BLACKBOARD
You can now create, store and view your personal timetable in
blackboard. A group of enthusiastic Computer Science students
were unhappy with the timetable site that was introduced last sum-
mer, and built a new, improved one. You can now create your time-
table based on the courses you are enrolled to, by study program
or individual. The software can export to iCal, CSV and PDF and
can be found in Blackboard.
STRESS ON THE THIRD FLOOR OF EWI
On the third fl oor people are working harder than ever. The
Bachelor and Master Electrical Engineering and the Master
Computer Engineering are about to be inspected. As such, our
whole education system has to be evaluated. Furthermore, a
new Bachelor program is in the pipeline. The plan is that up-
coming freshmen will start with a study program that is much
more project orientated.
RESEARCH INFORMATION FOR (BACHELOR) STUDENTS
Most Bachelor students have little idea of what goes on in the EWI
tower. That’s why the ETV, in cooperation with EWI Marketing,
is organising an event in May were every research group can talk
about the exciting research they are currently working on. Students
will hear more about this event as details are confi rmed.
De Facultaire StudentenraadWij als facultaire studentenraad, of kort FSR, zijn alweer een tijdje bezig en houden de faculteit goed in de gaten. We hebben in het begin
van het collegejaar een 10-puntenplan gemaakt met de 10-punten waar wij dit jaar extra aandacht aan geven. Dit doen wij vooral in onze
maandelijkse overlegvergadering met de decaan, maar ook in verschillende werkgroepen in de faculteit, waar onze mening gevraagd is. Een
paar van deze punten hebben wij hier even kort toegelicht.
INVOERING VAN DE HARDE KNIP EN HET BINDEND STU-
DIEADVIES.
Eerstejaars krijgen voor het eerst te maken met het bindend
studie advies, dat betekent dat ze tijdens hun eerste jaar mini-
maal 30 ECTS moeten halen en anders helaas niet verder mo-
gen met hun studie. Verder krijgt iedereen uit 2006 en later na
dit jaar te maken met de harde knip: eerst alle punten uit je ba-
chelor halen voor je verder mag met je master. Zowel rondom
het BSA als de harde knip zijn harde afspraken gemaakt door
de studentenraad en wij zullen er op toezien dat deze binnen
onze faculteit nageleefd worden.
UITSTRALING VAN DE FACULTEIT
We zijn dit jaar bezig de uitstraling van onze faculteit te verbeteren.
Zowel het hoofdgebouw als gebouw 35 stralen niet uit dat het EWI
gebouwen zijn, terwijl dat wel zou moeten. Heb jij werkbare ideeën
over de invulling van dit idee, laat dat dan weten!
DIPLOMAUITREIKINGEN
De ceremonie zoals deze op dit moment een aantal keer per jaar
plaatsvind wordt door velen als saai ervaren. We zoeken naar een
oplossing om de uitreikingen leuker te maken terwijl de studenten
die hun diploma ophalen toch de verdiende aandacht krijgen. Ook
ideeen hierover horen we graag van jullie.
Deze punten krijgen bijzondere aandacht van ons, maar ook alle andere problemen pakken wij natuurlijk aan. Dus heb jij een idee over een
van de bovengenoemde punten of kom je zelf andere problemen tegen? Laat dat dan weten via [email protected] of spreek een van ons
aan op de faculteit. Er is altijd wel iemand te vinden bij de ETV of CH, we zijn te herkennen aan de foto’s op de posters die door de hele
faculteit hangen. Vergeet ook niet je stem te laten horen bij de volgende verkiezingen van de FSR op 26 en 27 mei!
The faculty of Electrical Engineering, Mathematics and Computer Science is con-
stantly undergoing evaluation and change. Here are the most important goings-on
in our faculty at this moment.
Author: Frank Teunisse, Commissioner of Education
Maxwell 13.3 April 2010 9
Activities of the
Electrotechnische Vereeniging
Excursie CeBIT Auteur: Patrick Fuchs
Daar stond ik dan ‘s ochtends (4 uur!) met mijn
slaapogen de rest van de groep te zoeken. Toen de
bus eindelijk kwam kon ik mijn ogen niet gelo-
ven (misschien ook omdat ik m’n contactlenzen
niet in had). Dit was de meest luxe bus die ik
ooit had gezien. Misschien waren het de zacht
leren stoelen, of de vroege vertrektijd, en mis-
schien een combinatie van beiden, maar ik lag
wel heel snel te slapen achterin de bus.
Enkele uren later (weer wat opgefrist na een re-
delijke nachtrust) kwamen we aan op het “mes-
segelaende” waar CeBIT plaatsvindt. Even la-
ter stonden we met z’n allen in de eerste zaal.
Als eerste natuurlijk langs bij Alcatel-Lucent,
die onze kaartjes voor de entree van de CeBIT
beurs betaald hadden. Daarna viel de groep re-
delijk snel uit elkaar en ging iedereen naar de stands of hallen
waar ze het meest interesse in hadden. Omdat het beursterrein
zo groot was (te vergelijken met het centrale deel van de TU
Delft-campus in grootte), had ik sommige mensen de hele dag
niet gezien. Aangezien dit de eerste keer was dat ik überhaupt
op een ICT-beurs was (afgezien van de Games Convention
Leipzig, waar het zwaartepunt toch ergens anders ligt) was ik
maar wat gaan lopen, op zoek naar leuke gadgets en goodies.
Dan waren er ook nog de 3D-schermen en -technieken die al-
lemaal naast elkaar geshowd werden. Dit trok natuurlijk de
aandacht van veel mensen, maar weerhield ons er echter niet
van ook zelf deze brilletjes en beeldschermen te ge-
bruiken en te bekijken. Erover horen is een ding,
maar zelf de verschillen vooral waarnemen geeft
natuurlijk een heel ander (3D-) beeld. Nadat uwe
reporter zelf geïnterviewd was door een Russische
tv-zender (over een 3D-bril die ik aan het “testen”
was, vond ik het welletjes en gingen we door naar de
rest van de hallen.
Jammer genoeg ging de terugreis niet helemaal zon-
der problemen. We waren namelijk iemand kwijt-
geraakt op de CeBIT (waarschijnlijk onderweg naar
Hannover voor het avondeten, voor de details moet
je maar het Bestuur vragen) waardoor we meer dan
een uur vertraging opliepen. Achteraf was iedereen
gelukkig naar Delft gekomen en was het een zeer
geslaagde excursie.
Zelfs binnen de grote mensenmassa waren de ETV’ers vanouds goed te onderscheiden.
De nieuwste gadgets van Asus.
10 Maxwell 13.3 April 2010
De deelnemers van de Shell-excursie.
Excursie Shell Auteur: Felix Fikke
Op het moment dat voor de meeste studenten het eerste col-
lege begon, verzamelde op 16 maart een groep enthousiaste
ETV’ers zich op het station van Delft. De reis leidde naar
Moerdijk, alwaar we, samen met een delegatie uit Eindhoven,
een bezoek brachten aan Shell. De dag was georganiseerd en
werd begeleid door Alexander Bosman, een jonge elektro’er die
sinds 4 jaar voor Shell werkt. Zoals te verwachten van iemand
met deze achtergrond was de dag perfect georganiseerd.
We begonnen met een kleine rondvraag om vast te stellen wat
men over het werken bij Shell zou willen weten. Daarna vertel-
de Alexander over de diversiteit aan carrièremogelijkheden die
een multinational als Shell bood. Gevolgd door een presentatie
van Joke Driessen met een algemeen praatje over Shell Moer-
dijk. Deze fabriek is een van de grootste kraakfaciliteiten in
Europa. Hier wordt per jaar zo’n 4,5 miljoen ton ruwe aardolie
verwerkt tot Nafta, aardgas en andere producten.
Na de presentatie van Joke was het woord aan Wim de Wilt, de
‘global manager Electrical Engineering’ voor Shell. Hij vertelde
boeiend over zijn langdurige carrière bij Shell en beantwoorde
een hoop van de in het begin gestelde vragen. Hierna was er ge-
legenheid om tijdens een kleine lunch met één van de diverse
werknemers gesprekjes te voeren en informatie op te doen.
Na de lunch werden we voorzien van beschermende kleding,
waarna we een rondleiding door de fabriek kregen. Ons groepje
werd begeleid door een oud elektro’er uit Delft. Hij leidde ons
langs de kraakovens, transformatoren en andere grote machi-
nes. Erg interessant was natuurlijk de controlekamer waar de
stroomvoorziening geregeld word.
Als afsluiting kregen we nog een aantal zeer waardevolle tips
over het solliciteren en stage lopen bij Shell, waarna we ver-
moeid doch voldaan terugkeerden naar Delft.
Lunch lecture Technolution Author: Lucas van Dijk
On the 9th of March, Dr.ir. Paul van Haren visited our faculty
to tell something about Technolution. It was defi nitely a good
and informative lecture, which showed a bit about their design-
process for a system they’re developing for Philips. He started
off telling in what fi elds Technolution is active. Examples are
traffi c engineering (NS, ProRail, Rijkswaterstaat), High-Tech
(Philips, ASML, OCE and Agfa), and some other fi elds like
telecom and power engineering (KPN, Motorola, Eneco). It’s
nice to know that Technolution provides solutions to certain
problems with existing technology, they don’t do any research.
In Holland, there’s only one offi ce, and it’s located in Gouda.
If I may believe Dr.ir. Paul van Haren, there’s a nice atmos-
phere at their company, discussing a lot of new technologies
colleagues have discovered. People who fi nished an electrical
engineering, computer science, mathematics or physics study
are always required at Technolution as these kinds of people
often have a lot of knowlegde in one or more of the following
fi elds: electronics, embedded systems, programmeable logic or
information systems. In their experience, people from electri-
cal engineering can be used in all four fi elds, but who expected
anything else?
After this part, that was mainly about Technolution, he contin-
ued describing the design process of a machine which can sort
of look inside your body without making any cut. The screen
where the inside of the body will be displayed, is huge. Besides
only the body view it contains six ‘subscreens’ which all dis-
play data like patient information, machine position, and all
you can think of. The resolution of this screen was around
4000x3000 if I’m right, so a good graphicscard is required. It
also allows to add external devices to the screen through a DVI
connector. They designed their own DVI grabber which can be
put in a PCI-E 2.0 slot, and it actually does a lot more than
only grabbing the DVI signal: it also compresses and crops the
view a bit, because there wouldn’t be enough bandwidth avail-
able on a PCI-E 2.0 slot to grab 4 external devices at the same
time.
There was more on the lecture, but I hope this has given you
a good idea about the lecture. They defi nitely got my interest,
because it sounds like a great company which works on a lot of
different projects, in all sorts of fi elds.
Maxwell 13.3 April 2010 11
Eestec LC Delft Author: Jean-paul Schouwstra
The ETV is a member of the international organization EESTEC, where it is known as EESTEC LC (Local Committee)
Delft. The main objective of EESTEC is to improve contacts between electrical engineering students from all over Europe.
This is achieved by organizing exchanges, an annual conference, workshops and other great activities. The principle of
EESTEC is simple: if you arrange your own trip to your destination location, your accommodation, food and program will
be arranged for you. All of the 40 LC’s organize an exchange at least once every two year. Activities that participating stu-
dents can expect are cultural activities, workshops, lectures and most importantly multicultural integration between the
students from all the different countries. Do you like this principle and want to stay up to date once a month on upcoming
activities and exchanges? Contact [email protected] and subscribe for the Eestec-newsletter!
For this year an enthusiast committee of the ETV started preparing an exchange to Delft. The theme of this exchange is
“Size Does Matter”. To prove this we have a very large committee, especially compared to other years. This committee will
ensure that the week of June the 7th will be an incredible, unforgettable week for the international guests as well as for the
Delft students participating.
Wouldn’t you like to be brought in contact with a nice foreign electrical engineering student? Well this is your chance. For
our exchange to Delft we are still looking for some hosts to provide sleeping places for international students for one week.
So if YOU still have some free square meters, please let us know by e-mailing [email protected]! In return you’ll be
able to go to some GREAT parties and get the chance to “keer” an exchange student.
An impression of the last EESTEC-exchange in Delft.
12 Maxwell 13.3 April 2010
Since July 1, 2004 C2000 replaced the an-
alog communication networks of Dutch
emergency services. Nowadays the police
forces, fire departments, ambulances and
the Koninklijke Marechausee (a force pro-
viding military police and civil police du-
ties) all use the C2000 system for their
communication. Recently, this system
caused quite some unrest in the Nether-
lands, due to some possible failures of the
system during some major events. In this
article the concerns that are associated
with C2000, will be pointed out.
Before we look at the issues that arose
with C2000, we first address the reasons
for the transfer to C2000. After that a
brief technical analysis of the C2000 sys-
tem will be given. Finally, the problems
that arose in using C2000 will be pointed
out.
From analog to digitalIn the mid nineties there were several de-
velopments, which caused an understand-
ing among the Dutch emergency services
that it was necessary to replace the old
analog communications networks. Be-
sides the general tendency towards the
digital era, one of the main developments
was the insight that the analog system
was not providing enough capacity and
lacked in security. Besides that, the ana-
log system did not provide the possibility
for the different emergency services, like
the police, fire squad and ambulances, to
communicate with each other. Both the
evaluation from the fire at Volendam dur-
ing the turn of the year 2000 to 2001 and
the evaluation of the fireworks explosion
“We need immediate assistance at the Mekelweg 4 in Delft. Over”. “Unit Echo Tango Victor is on
route. Over and out”. Seems like a random example of a possible communication message between a
mobile unit and the radio control room of one of the emergency services. It holds in general that com-
munication is of vital importance in the operation of emergency services. Therefore it is easily under-
stood that the communication network of the emergency services in the Netherlands, C2000, recently
received a lot of attention due to some possible failures of the system.
C2000Life saving communication?
Author: Stephan Groot
Maxwell 13.3 April 2010 13
in Enschede, confirmed these issues some
years later.
In September 2004 the C2000 was intro-
duced. It took three years before the last
police squad, the one of the region Am-
sterdam-Amstelland, shut their analog
network down and switched to C2000.
This completed the nationwide migra-
tion towards the digital C2000. C2000
belongs to the class of Private Mobile Ra-
dio (PMR) systems, which have different
requirements than the public communi-
cation networks. The specific demands of
users of PMR systems will be addressed
later. These demands cannot be satis-
fied by conventional system like GSM,
GSM-R and DECT. Therefore a standard,
especially suitable for PMR systems, like
C2000, is developed.
TETRAThe C2000 communication network
makes use of the standard TETRA,
which is an abbreviation for TErrestrial
Trunked RAdio. Trunking is a word bor-
rowed from the telephone system to de-
scribe a system, in which a large number
of users share a much smaller number of
communication paths. Traditionally, this
was related to a wire (the trunk) that was
assigned to a telephone user upon mak-
ing a call. In the case of terrestrial radio
the scarce resource is bandwidth. If we
relate this for example to the police in a
certain region, where there are at a given
time quite a number of police officers on
duty, who all need to stay in contact with
a dispatcher. If each officer would use an
exclusive channel, the available frequen-
cy band would soon run out of possible
channels. Moreover, this is not a very ef-
ficient way to assign radio channels, since
each channel would be idle most of the
time.
In a trunked radio system the total num-
bers of users are divided into groups. Im-
agine waiting with a group of friends for a
table at a crowded restaurant. You go up
to the hostess and give her your name,
and she puts it on a list with a bunch of
other names. If all the tables already have
people at them, you wait. When a table is
ready the hostess announces your name
over the speaker and you and your friends
follow her to the table she selected for you
(probably the first one that became avail-
able). This principle is also applicable
to the kind of trunking used in TETRA.
When a user of one group wants to talk to
a user of another group, he has to request
a channel assignment from a controller
first. The controller will check if there is a
channel free. If all the channels are occu-
pied, the controller makes you wait until
a channel is free, and then publicly an-
nounces your talk group and the assigned
channel. User A and the intended receiv-
ing users then switch to that channel and
can communicate with each other.
Trunking can be applied to both a part
of the conversation or to the entire con-
versation. The first is implemented in
TETRA. In this case a conversation that
takes place over several transmissions
may actually occur on several different ra-
dio channels because the controller may
assign a new channel every time someone
presses their push-to-talk button. This
is the most efficient way to share radio
channels, since other people can use the
channel during pauses in the conversa-
tion, but it is also what makes it more
difficult for a normal scanner to listen in.
At this point, one can wonder why use
a complete different standard instead of
proven technology like GSM. The rea-
son for this is the specific requirements
that communication networks for emer-
gency services demand. The traditional
techniques, like GSM, have certain is-
sues, discussed here, which are solved in
C2000.
GSM limits the amount of subscribers
in a group to 1024 and the amount of
dispatchers to 5. In TETRA no restric-
tions on group size are defined, but a
group can deal with at most 30 dispatch-
ers. Especially the relative small number
of possible dispatchers in GSM is a limi-
tation, since one of the objectives of the
network is to enable different emergen-
cy services to communicate with each
other. This requires many dispatchers.
A very effective way of informing mem-
bers of a group in case of an emergency
is trough sending a short data message
to each member of the group. This func-
tion is defined in TETRA, but is not
available in GSM. Of course it is pos-
sible to send a SMS in GSM to each
member of the group individually. How-
ever, this is significantly slower than the
service TETRA provides. Besides that,
since SMS data files run over the signal-
ing channel of GSM, this will put a sig-
nificant additional load on the signaling
channel, when used extensively.
Another rather important difference
origins in the way how both handle
call priorities. Call priorities are speci-
fied for both TETRA and GSM. How-
ever, TETRA is more flexible in this
case, because it can provide different
priorities to a user or group. Further-
more, in TETRA more priority classes
are specified. Another rather impor-
tant feature of TETRA that is not in-
corporated in GSM is the speech item
priority. In for example emergency
situations it can be important that
a group leader can get a speech item
and force the other subscribers in the
group to listen. This feature can be
important to achieve organized com-
munications, which can be critical in
an emergency situation.
When designing a communication
network for the emergency service
the call set-up time is of huge impor-
tance. In C2000 it is required that a
call can be set-up within 0.5 seconds.
With some efforts the call set-up time
in GSM can only be reduced to one
second.
14 Maxwell 13.3 April 2010
The security of the network is of ex-
treme importance for the intended
users of C2000. The security should
ensure that someone not belonging to
a certain group is unable to listen to
the communication, but also is un-
able to disturb the communication.
In TETRA three types of security
measures are implemented. First, au-
thentication ensures that only users
with a valid key can use the network.
Moreover, the call information, the
control information and the identity
of the users is encrypted, by a so-
called Air Interface Encryption (AIE)
protocol. This ensures that a user can
be tracked by following the signal-
ing messages on the control channel.
Last, encryption between the end-
points of communication is provided.
End-to-end encryption (e2ee) prevents
eavesdropping by encrypting the in-
formation on the traffic channel. The
difference with AIE is that the infor-
mation on the control channel isn’t
encrypted.
IssuesIn 2009 there were several incidents,
which caused some distrust in the C2000
system. In February a passenger flight of
Turkish Airlines crashed near Schiphol
Airport. During the rescue the system had
signs of overloading. For example many
people couldn’t get a proper connection
using their radio. Afterwards, the Inspec-
tion for Public Order and Safety conclud-
ed that the available 11 groups were too
little for the vast number of emergency
services on site. Another, and maybe one
of the most well-known, example of the
disfunctioning of C2000 is during the
riots on the beach of Hoek van Holland.
When the situation escalated, again, the
system showed signs of overloading. The
police men on site couldn’t communicate
with each other for some time.
As a result of these incident, the Depart-
ment of Interior Affairs has investigated
the operation of C2000. One of the out-
comes of this investigation is that there
are some locations in the Netherlands,
where the coverage of C2000 is not suf-
ficient. At these locations the link tends
to switch between two different base sta-
tions, hence drastically deteriorating the
connection. These locations are collected
in a so-called DIPP list (Dekkings Is-
sue Prioritering Procedure, in short this
means a collection of the important lo-
cations, where coverage problems arise).
This DIPP list also includes Pernis, where
the Dutch petrochemical industry is con-
centrated, Borssele, the location of a nu-
clear power plant and Vught, the location
of an extra secure prison. Recently was
announced the coverage of 60 high-risk
locations like these will be optimized
within two years.
Although some of the technical short-
comings of C2000 are recognized, the
influence of the user on the proper func-
tioning of C2000 also has to be taken
into account. For example when evalu-
ating the problems that arose during the
plane crash in February, it was concluded
emergency people were using their radio
for non-essential communication too. If
the radio was only used for the important
messages, the overloading of the network
could have been prevented. The failures
of the system, during the riots at Hoek
van Holland, but also during the attack
on Queen’s Day 2009 were ascribed to
the fact that there were too many us-
ers in the same group. The director of
the region Rotterdam-Rijnmond, Don
Berghuis, who was appointed for the re-
search, states that many emergency serv-
ice people all use the same group. There-
fore the system seems overloaded. In fact
this isn’t the case, but because there are
many users in one group, individual users
need to wait longer before they can speak.
When in critical situations people tend to
speak at the same time, this gives prob-
lems. Berghuis also added that this isn’t
a specific telecommunications problem.
If ten people are talking to each other in
person, communication also won’t work
if they talk at the same time. To solve this
problem, the users of C2000 are trained
again to work properly with the system.
ConclusionMore than a decade ago the Dutch emer-
gency services switched from an analog
communication network to a digital one:
C2000. Theoretically, this new commu-
nication network provides the emergency
services with state-of-the-art commu-
nication, providing in all their specific
demands. However, when using C2000
some issues arose. Especially during ma-
jor incidents, the system showed signs of
overloading. Investigations showed that
there exist both technical as user related
reasons for these issues. To overcome
the technical problem of too little base
stations at critical locations, the coming
years these base stations will be installed.
The users of C2000 are once again
trained, in order for them to work prop-
erly with the system. Maybe, these two
actions will finally enable the emergency
services to use their life saving communi-
cation system!
References:[1] Terrestrial Trunked Radio – TETRA, P.
Stavroulakis, Springer, 2007.
[2] Eindrapportage expertgroep C2000 –
Ministerie van Binnenlandse Zaken en
Koninkrijksrelaties, definitieve versie, 22
december 2009.
Maxwell 13.2 January 2010 15
HIER INVOEGEN:PAGINA15SMIT.PDF
16 Maxwell 13.3 April 2010
The list of brain diseases is very long: Alzheimer's disease, Par-
kinson's disease, Tinnitus and mental disorders (e.g. clinical
depression or schizophrenia) are just a few of the well known
examples. All these diseases have major impact on the quality
of life of the patients. Existing treatments have only limited
effectiveness.
During my Master thesis I have looked into the design of a neu-
ral stimulator: a device which is able to directly infl uence the
functionality of the brain using electrical stimulation. It turned
Controlling the brainHow electrical stimulation can be used as an
effective treatment for many brain disorders.
out this leads to a very exciting research
fi eld with many challenges which need to be
solved; both in the medical as well as in the
electrical domain.
In this article an introduction will be given
in the fi eld of neural stimulation. The focus
will be on how the brain works from an elec-
trical point of view and how it is possible to
infl uence this functionality using stimula-
tion. Also an overview is given for the most
important innovative features our design is
providing compared to existing stimulators.
Why is brain stimulation effective?Throughout history medicine has been going
through an extensive development. Prehis-
toric medicine consisted of a combination of
very basic drugs (extracted from plants and
herbs) and treatments consisting of ceremo-
nies carried out by shamans using supernat-
ural powers and objects (charms, spells, am-
ulets, etc). During the course of history more
and more knowledge was gained about the
use of drugs for treating diseases. From the
19th century on medicine has gone through
a revolution thanks to advances in for exam-
Figure 1: Photo’s of the surgical procedure of implanting the electrodes for neural stimulation in the brain. In the right picture the electrode lead is inserted
in the brain (pictures courtesy of dr. D. de Ridder)
Our brain is the center of our nervous system. Literally everything we do,
from eating an apple to solving a Schrödinger equation, is controlled by
it. Usually we don’t give much thought to the fact our brain is so utterly
important, but imagine something starts to go drastically wrong inside the
brain. Not being able to solve a Schrödinger equation might not be a big
problem for the majority of people, but eating an apple is.
ple chemistry and today drugs form an utterly important ingre-
dient of medical treatments.
However it is more and more realized that drugs alone can-
not cure all diseases effi ciently enough. Most drugs suffer from
unwanted side effects and the spatial selectivity of drugs is usu-
ally low: it will have effects on the whole body, while usually a
disease is limited to only certain parts. Another form of treat-
ment consists of electromagnetic stimulation of the body. Cells
use electromagnetic signals to operate. These signals can be af-
Author: Marijn van Dongen
Maxwell 13.3 April 2010 17
fected by artificially generated electromagnetic signals in order
to establish a desired effect.
Probably the most well known form of stimulation is the pace-
maker: a stimulator for the heart muscle. This particular tech-
nique has gone through an extensive development as well. As
early as 1820 Richard Reece described in his 'Medical Guide' a
method for stimulating the heart. A metal rod was inserted in
the esophagus and connected to a voltaic
cell, while another rod was pushed to
the chest by the physician. In this
way a 'manual' pacemaker was
comprised as depicted in Figure
2. Nowadays pacemakers are
very sophisticated implanta-
ble stimulators: they record
heart activity and based
on this activity the de-
vice can decide when
and how the heart
needs to be stimu-
lated.
The brain essentially
works in a similar man-
ner as the heart muscle:
neural cells communicate
by 'activating' each other
using electrical impulses.
This means that the activity of the brain can be affected by
means of electrical stimulation as well. As shown in the in-
troduction, many diseases find their origin in abnormal brain
functionality. It is therefore in principle possible to eliminate
this unwanted activity by using brain stimulation. This can be
done by implanting electrodes connected to a stimulator, which
can deliver stimulation pulses to the tissue.
In this way the functionality of the brain is directly influenced.
This results in a much more selective treatment. Side effects
are expected to be much smaller, thereby increasing the effec-
tiveness.
The development of neural stimulators is still in a relatively
early stage: in some way current stimulators are comparable
to the first pacemaker from 1820. Most stimulators have lim-
ited implantability as depicted in Figure 3. Furthermore they do
not incorporate any feedback: they simply stimulate the neural
tissue using a fixed stimulation pattern, without recording the
activity in order to adjust the stimulation if required.
How does neural stimulation work?Before we can start with the
design of a neural stimula-
tor, first a closer look is tak-
en at the physical principles
underlying electrical stimu-
lation. It is required to know
what exactly happens in the
tissue when electrical energy
is transferred to an electrode
implanted in the brain.
The brain consists of tens
of billions neural cells called
neurons. A neuron has a cell
body (soma) with a nucleus,
as depicted in Figure 4. The
inputs of the cell are called
dendrites and its output is called an axon. Input voltage pulses
coming from the dendrites (sometimes thousands of dendrites
are connected to a single neuron) are processed in the nucleus
and this can result in an output pulse in the axon. These pulses
form the fundamental mechanism for all actions in the central
nervous system.
The voltage pulses are transferred over the cell membrane,
which encloses the neurons and axons. Within this membrane
there are ion channels through which different types of ions
(mainly potassium, sodium and chloride) can flow. Due to a
concentration gradient of the ions between the inside and out-
side of the membrane, there is a continuous ion flow through
this membrane.
The flow of ions is determined by the conductance of the
channels, which turns out to be dependent on the membrane
voltage. This will eventually lead to a dynamic equilibrium in
which ions will flow with a constant rate, yielding a resting
potential of the membrane.
Changing the membrane voltage, will change the ion flow ac-
cordingly. When the membrane voltage is raised above a par-
ticular threshold, the membrane is 'activated' and a process is
triggered, which will result in a voltage pulse over the mem-
brane. This pulse, often referred to as an action potential, will
propagate over the axon towards other cells.
When an electrode is now placed in the tissue, it is possible
to change the outer membrane potential by applying a particu-
lar voltage to this electrode. In this way it is possible to
Figure 2: Electrical stimulation anno
1820 according to Richard Reece. A
manually controlled non implantable
device. State of the art neural stimulators
are surprisingly similar to this ancient
medical device.
Figure 3: Existing stimulators have
limited implantability. The stimulator
can only be implanted in the chest,
while subcutaneous wires lead to
the electrodes in the brain (Image
courtesy of Medronic inc.)
18 Maxwell 13.3 April 2010
change the membrane voltage and therefore action potentials
can be artificially generated or blocked. In this way the func-
tionality of the brain can be directly influenced.
Therefore neural stimulation is from an electrical point of view
nothing more than elevating the tissue potential (the outer
membrane potential) up to a particular threshold. Electrical
modeling of the electrode-tissue interface shows a capacitive
nature. This means that elevating the tissue potential above
a particular threshold corresponds to injecting a particular
amount of charge in the tissue. This observation has been a
critical starting point in the project.
Design of the stimulatorThe fact that charge is the fundamental electrical quantity for
stimulation can be used to design an effective stimulator. How-
ever another important design aspect is safety. When applied
in the correct way, stimulation can lead to beneficial results in
the brain. However, stimulation can also lead to damage to the
tissue when certain requirements of the stimulation waveform
are not met.
One of the basic safety constraints is that no net charge can be
injected into the tissue. Charge at the electrode tissue interface
will lead to electrolysis, which is harmful to the tissue. This
means that after a stimulation pulse some mechanism must
make sure that the injected charge is removed from the tissue
again. One fast way to do this is by applying a second, negative
pulse with the same charge contents.
Taking these aspects in mind, one of the first choices for the
design of a stimulator is the electrical quantity which is used
to inject the electrical energy into the tissue. Most of the exist-
ing stimulators use constant value current sources to stimulate
the tissue, as illustrated in Figure 5a. In this way the amount
of charge injected can be quite easily controlled: by enabling a
constant current for a particular amount of time, the charge is
defined as Q=It. There are however a few important drawbacks
related to current sources:
High power consumption
For charge cancellation, the value of the current source must be
equal during the stimulation phase as well as the charge cancel-
lation phase. To create accurately matched current sources, a
lot of power is required. This will increase the size of the bat-
tery. Current stimulators have limited implantability, because
of the size constraints which are determined by the battery. In-
creasing the power efficiency is therefore very important.
Waveform flexibility
One major drawback of the existing stimulators is the limited
adjustability of the waveform. To be able to control the charge
easily, usually only square shaped pulses can be injected. Neu-
ral tissue shows an amazing adaptability to stimulation. This
means that due to the fixed stimulation pattern the tissue will
gradually habituate to the stimulation pattern, which means
the symptoms of the disease return.
It is expected that habituation can be reduced when it is pos-
sible to generate varying stimulation waveforms. Furthermore a
wide waveform choice can result in more effective stimulation.
Examples include sinusoidal or triangular shapes, asymmetric
stimulation and a wide variety of burst stimulation. Using cur-
rent sources it is hard to realize this: the charge is not controlled
as easy anymore compared to the constant current approach.
Based on the two observations above and because current is
not a fundamental quantity, it was decided to look into alterna-
tive electrical quantities to steer the tissue. It was shown that a
voltage source (see Figure 5b) yields a much better approach: it
is much more power efficient and it is easy to make a variable
voltage source for waveform flexibility.
Figure 4: Structure of a typical neuron and a detail of a cell membrane (Image courtesy of Mariana Ruiz and Quasar Jarosz, modified under the Creative
Commons BY-SA license.)
Maxwell 13.3 April 2010 19
The drawback of a voltage source approach is that the charge is
not easily controlled anymore, since the current is determined
by the tissue impedance, which is very time variant. Therefore
it was decided to include a current feedback loop to keep track
of the charge injected. In this way it is possible to use any wave-
form, while still ensuring charge cancellation.
A new type of system architecture was proposed using indirect
current feedback. This also implied the use of an additional
voltage feedback loop to control the tissue voltage. This dou-
ble feedback architecture has been implemented in Amis 0.35μ
high voltage technology (stimulation voltages used are up to
20V). This principle is depicted in Figure 6.
The implementation details of the system are beyond the scope
of this article, but probably the most challenging part of the
design was the current feedback loop in which an integrator
needed to determine the charge injected based on the stimula-
tion current. This integrator was required to handle a dynamic
range of several decades of magnitude.
ResultsSimulation results on circuit level show the feasability of the
system. An example of a stimulation waveform generated by
the circuit (1kHz, 2V amplitude and 7V offset) was injected
into the tissue (modeled as a parallel combination of a resistor
of 10k and a capacitor of 75nF). The charge threshold was set
to 171nC. The transient simulation result of the tissue volt-
age is depicted in Figure 7. As can be seen, the tissue is first
charged to about -2.3V during the first (negative) voltage pulse.
Subsequently a positive pulse is injected to remove the charge
at the tissue. As can be seen the resulting tissue voltage is very
close to zero, indicating the charge metering technique is work-
ing properly.
One of the design goals was to have a very low power consump-
tion to increase the implantability of the device. Active power
consumption is very dependent on the waveform used. There-
fore it is hard to quantify the active power consumption or ef-
ficiency of the design.
The quiescent power consumption is however as low as 15μW.
Most of this power is burnt in the voltage feedback network
to bias the gain stage here. Furthermore the switches require
some bias to generate a floating voltage source controlling the
switches. The quiescent power consumption is to the best of
our knowledge among the lowest values for quiescent power
consumption reported until now.
An important safety performance parameter is charge mis-
match. In the waveform from Figure 7 the remaining charge
was 1.5nC, corresponding to about 1%. About 50% of this
mismatch is due to discharge of the tissue in the inter-pulse
delay because of the finite off resistance of the switches in the
switch array. This mismatch can therefore easily be removed
when the inter-pulse delay is chosen to be shorter. Further an-
other 40% of the charge imbalance is due to inaccuracies in the
implementation of the integrator. When these inaccuracies
Figure 5: Illustration of two possible fundamental stimulation architectures
Figure 6: The principle of indirect current feedback with direct voltage
feedback, which is used as the fundamental system architecture.
a. Current steered b. Voltage steered
0 1 2 3 4 5 6 7 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8
Time (s)
Tis
sue
volta
ge (
V)
Figure 7: Tissue voltage during excitation with a sinusoidal waveform.
The final tissue voltage is close to 0V, illustrating the feasibility of the
system.
20 Maxwell 13.3 April 2010
are improved, the charge mismatch will become 0.1%. Any
remaining charge can be discharged from the tissue by short
circuiting the tissue electrodes using the switch array if needed.
Another safety parameter is the ability to handle the large spread
in stimulation and tissue parameters. The system is working
for any combination in tissue parameters ranging from 1k <
R < 100k and 10nF < C < 100μF. Furthermore the system is
also working over all process corners and process mismatches,
preserving charge cancellation.
Because of the chosen architecture there are endless possibili-
ties for waveform adjustments. In principle any waveform can
be used: the charge metering mechanism will keep track of the
charge injected in the tissue. It is therefore possible to use tonic
stimulation, burst stimulation, asymmetric stimulation, sub-
threshold prepulses, excitatory and inhibitory stimulation, etc.
To illustrate this two waveforms for both tonic and burst stimu-
lation are depicted in Figure 8. This figure illustrates the charge
cancellation mechanism is working for a wide variety of shapes,
since the final voltage is very close to zero.
ConclusionsIt has been shown that electrical stimulation of the brain opens
up a huge field of treatments for a wide variety of diseases. Be-
cause of the highly selective properties of neural stimulation it
can offer more effective treatments in many cases compared to
the use of drugs.
The physical principles underlying electrical stimulation of
neurons have been investigated in detail. It was shown that
from a electrical point of view the stimulation of neural tis-
sue is equivalent to raising the extracellular potential up to a
particular threshold value. Because of the capacative nature of
the electrode tissue interface, this is equivalent to injecting a
particular amount of charge into the tissue.
Taking this principle in mind, a fundamental new system ar-
chitecture was designed and implemented on the circuit level.
Simulation results confirm the feasibility of the system. Fur-
thermore the system has a very versatile character (endless
waveform possibilities), very low power consumption, while
safety is still assured.
This design opens the way to more efficient neural stimulation
treatment methods. Still, there is much room for improvement.
System blocks like the external communication, power man-
agement and an automated feedback loop (stimulation based
on recording neural activity) still need to be designed. There-
fore this project will continue in order to design the remaining
system blocks to finally come up with a completely new and
revolutionary neural stimulator.
The project is carried out in cooperation with many univer-
sity hospitals, especially the BRAI2N clinic in the University
Hospital of Antwerp (UZA). The multidisciplinary approach of
this project is something I have very much enjoyed during the
course of the project.
If you have questions related to this article or you are interested
in the details of the electronic implementation, you are very wel-
come to contact us. Within our group we are also always looking for
motivated students who are willing to contribute to this challeng-
ing project or other projects in the biomedical field. The contact
details for me and my supervisor Wouter Serdijn are as follows:
0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8
time (s)
Tis
sue
volta
ge (
V)
0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8
time (s)
Tis
sue
volta
ge (
V)
0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8
time (s)
Tis
sue
volta
ge (
V)
0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8
time (s)
Tis
sue
volta
ge (
V)
Figure 8: Illustration of the endless waveform shape possibilities. The tissue voltage is showed for two different tonic and burst stimulation waveforms,
while charge cancellation is preserved.
Biomedical electronics group, Elca Research Laboratory,
Mekelweg 4, 2628CD Delft
Marijn van Dongen dr. ir. Wouter Serdijn
Room HB 18.030 Room HB 18.310
Maxwell 13.2 January 2010 21
HIER INVOEGEN:PAGINA21TNO.PDF
22 Maxwell 13.3 Aprli 2010
AvionicsThe main contributor to innovation in aviation
The need for a specialization of Electrical En-
gineers in the area of Avionics was already rec-
ognized in 1976 by prof. van Oosterom from
the Faculty of Aerospace Engineering at Delft
University of Technology (at that time still TH
Delft). To address the identifi ed need, in 1979
the Faculty of Electrical Engineering started a
special program which aimed to educate inter-
ested students with a background in Electrical
Engineering in the area of Avionics. At present,
it is an interfaculty specialization profi le. Al-
though having been restructured several times
in the past 30 years, the core philosophy of the
program remains the same:
• Use a set of coherent courses from the Fac-
ulty of Aerospace Engineering and the Faculty
of Electrical Engineering to provide the students
with the background needed to successfully
perform an M.Sc. research project in the area of
Avionics.
• Focus the Avionics research on a number
of internationally recognized challenges and
pursue them through international cooperation.
• Maintain a state-of-the-art research infra-
structure to provide an inspiring learning and
research environment.
The frequent reception of best-student paper
awards at the largest international Avionics con-
ference (organized jointly by IEEE and AIAA)
and the contribution of many former EWI Avi-
onics students in highly visible international
research programs (such as the NASA Aviation
Safety Program) prove that this approach is very
successful. To provide EWI students that are
fascinated by aviation a better idea of how they
can obtain the required background for a future
career in the fi eld of Avionics, this article starts
with a brief overview of today’s challenges. Us-
ing examples from previous and current inter-
national research projects, it is illustrated how
M.Sc. and Ph.D. students and researchers from
EWI have contributed to defi ning the future
state-of-the-art in Avionics.
Figure 1 (background): Cockpit of EWI simulator.
Maxwell 13.3 April 2010 23
Avionics refers to Electronic systems used in Aviation, and the word itself is a blend of Aviation and
Electronics. Avionics are not only essential for today’s commercial and military aircraft to fl y, but also
enable their integration into the overall traffi c management system. For safety critical applications such
as navigation, aviation imposes requirements and constraints on the electronics system which are far
more stringent than for example in the consumer electronics domain. This has an impact on the design
of the Avionics.
Author: Dr.ir. Erik Theunissen
Innovation in aviationWhen comparing a Boeing 747 that was built in
1969 with a Boeing 747 built in 2010, the biggest
leap in capabilities can be attributed to Avionics.
Engines have become far more effi cient through
the use of Fully Autonomous Digital Engine
Control (FADEC). Safety has increased through
the introduction of the Enhanced Ground Prox-
imity Warning System (EGPWS) and the Traffi c
alert and Collision Avoidance System (TCAS).
Through better integration and automation,
the number of crew members that is required
to operate the aircraft has been reduced to two.
The instrument Landing System (ILS) provides
the capability to land under zero visibility con-
ditions, and the Flight Management System
(FMS) allows fuel optimized path to be fl own.
Indeed, in the past 40 years, a large part of the
innovation in the aviation domain was the re-
sult of developments in the area of Avionics.
The reason for this leap is that for Avionics the
enablers lie in the electronics domain and hence
many capabilities advance with the speed of
Moore’s law. Also today, the challenge for Avi-
onics is to benefi t from this advantage when ad-
dressing the issues aviation is confronted with.
The following four issues are internationally re-
garded as top priorities:
1. Reduction in the environmental burden
caused by aviation
2. Further increase in safety, both in the air
and on the airport
3. Further increase in airport availability (inde-
pendent of the actual weather)
4. Integration of unmanned aircraft into con-
trolled airspace
Environment: There is still ample potential
of Avionics to reduce the burden of Aviation
on the environment. The theme of the IEEE/
AIAA 2010 Digital Avionics Systems Confer-
ence (DASC) is ‘Greening Aviation’. Whereas
today’s instrument landing system only
24 Maxwell 13.3 April 2010
provides guidance for straight-in approaches, future Avion-
ics systems, enabled by more accurate and reliable sensors,
will support precision guidance along curved paths. This
provides the opportunity to circumvent noise sensitive ar-
eas that lie below the current approach paths. Through op-
timized navigation in the vertical dimension, the current
transitions between level and descending path segments
during an approach can be minimized or even eliminated,
allowing a significant reduction in the required engine
thrust and thus reducing emissions. To safely fly these more
complex approaches, advanced cockpit displays are required
to provide the pilot with the information needed to assess
whether the aircraft stays free of any hazards.
Safety: During the past thirty years, the major improve-
ments in the safety of commercial aviation were achieved
through the extrapolation of sensor-based information. To
reduce midair collisions, TCAS interrogates the transpond-
ers on board of other aircraft to determine whether there is
a collision hazard. Likewise, GPWS uses data from the radar
altimeter to determine whether the aircraft inadvertently is
getting close to the terrain. In the past ten years, the devel-
opments in the area of compact data storage have allowed
GPWS to be augmented with a warning system that relies
on the use of a worldwide terrain database and GPS-based
position information. These databases can also serve to pro-
vide the pilot with a synthetic view of the environment, in-
dependent of visibility conditions. Before actual operational
credit can be taken for this capability, a concept is needed
that timely detects hazardous discrepancies between the da-
tabase and the real world.
Availability: Whereas large airports such as
Schiphol are equipped with guidance systems that
allow properly equipped aircraft to land under
zero visibility conditions, many smaller airports
do not have such equipment. Reasons comprise
cost of the required infrastructure but also loca-
tion. In a mountainous environment, today’s ILS
cannot be deployed because of undesired signal
reflections. A navigation system that meets or
exceeds the accuracy, continuity and integrity re-
quirements of today’s (ground-based) ILS without
requiring significant infrastructure on the ground
is one of the top priorities for Avionics research.
Unmanned aircraft: Recent guidance from the
U.S. Federal Aviation Administration (FAA) sug-
gests that self separation needs to be a component
of an Unmanned Aircraft System (UAS) Sense
and Avoid solution in order for UAS to behave similarly to
manned aircraft. Conceptually, UAS self separation is the
protective (or conflict avoidance) method that precludes a
threat aircraft from ever triggering a time-critical collision
avoidance maneuver. To realize such a self-separation ca-
pability requires an Avionics system comprising sensors to
detect the collision hazards, algorithms to filter out observa-
tion noise and compute solutions for those threats that are
identified as real ones, and user interface concepts to allow
an operator to monitor the situation.
Avionics research @EWIWithin the EWI Avionics research program, specific chal-
lenges related to the aforementioned issues are being ad-
dressed. To benefit from similarities between the differ-
ent challenges, the research is structured along common
themes such as integrity, path optimization and human
machine interfaces. The expertise at EWI covers hardware,
software, algorithms, user-interfaces and system interfaces,
allowing the multidisciplinary challenges that are typical
for Avionics to be properly addressed. The EWI Avionics re-
search has an excellent reputation on an international level.
In 1998 the Faculty of Electrical Engineering (EE) was in-
vited by industry to join them in an effort to compete for re-
search in the context of the NASA Aviation Safety Program.
As part of a team lead by Avionics manufacturer Rockwell
Collins, synthetic vision system prototypes were developed,
integrated and first flight tested in 2000. Based on the re-
sults, Avionics researchers from EWI further refined these
prototypes and in 2001 the resulting system was success-
Figure 2: Cockpit of NASA 757 with EWI SVS.
Maxwell 13.3 Aprli 2010 25
fully evaluated in the NASA Boeing 757 during
the Eagle-Vail flight tests (figure 2). In August
and September 2001 more than 100 continu-
ous descending curved approaches were flown
in a terrain challenged environment.
One of the Avionics students that became in-
volved in the research in 1999 during his M.Sc.
project stayed on as a part-time researcher and
was one of the three engineers that Rockwell
Collins allocated to support this NASA flight-
test program. For the contribution to the over-
all NASA Aviation Safety program the Turning
Goals Into Reality Award was received.
The system test flown in 2001 relied on the
use of a digital terrain database to provide the pilot with
information about the terrain hazards. The Achilles-heel
of any database-oriented system is that errors in the data-
base can cause terrain hazards present in the flight path of
the aircraft not to be depicted. To enable timely detection
of hazardous discrepancies between the real environment
and the synthetic environment, the prototype system was
extended with the capability to integrate a real-time sen-
sor images. In 2004, four EWI Avionics researchers partici-
pated in the NASA flighttests during which this system was
evaluated (figure 3).
As indicated earlier, part of the focus is on unmanned sys-
tems. Several Avionics students performed their research in
this area on topics such as automatic landing systems, air-
space integration and the use of networks to achieve control
from geographically separated locations. Prototypes of UAV
mission management stations developed in a joint project
between EWI and the Royal Netherlands Naval College are
used by the Royal Netherlands Air Force to explore new
concepts of operation using mission level simulations.
The amount of rewards that has been received indicates
that the quality of both the M.Sc. projects and the overall
research is internationally recognized. In 2005, M.Sc. stu-
dent de Vries received the IEEE/AIAA Best Student Paper
Award at the DASC for his Avionics research project. The
following year, M.Sc. student ‘t Hart received this award for
his M.Sc. project that focused on UAV control. In the past
five years, the Avionics research was awarded with a total
of fourteen international awards, two of which (in 2007 and
2008) were ‘Best of Conference’.
Avionics education @EWIThe EWI Avionics program prepares interested students
with the background required to become an Avionics En-
gineer. This involves both courses such as ET4138 which
focus on the Avionics systems and courses at Aerospace
Engineering dealing with aircraft performance (AE4-220ET)
and flight dynamics (AE3-302)
The Avionics program is supported through an excellent
research infrastructure. Room 20.320 on the 20th floor
houses a research UAV operator station with which new
concepts for mission management can be evaluated. Room
20.070 houses a research flight deck, equipped with a pro-
grammable Electronic Flight Instrument System, a Flight
Management System, simulation of all required sensor
systems, and a projection system to simulate the view out
of the cockpit (figure 1). This flightdeck is used both for
research projects and in the avionics education program
(ET4244).
Avionics is a fascinating area and quite frequently students
that became involved with a particular topic during their
M.Sc. project continued to work in this area. Some during
a subsequent PhD research project, some as a researcher
at EWI or another University such as Ohio, and others at
institutes such as NLR, organisations such as LVNL and
companies such as ADSE and Rockwell Collins.
If you are a student at EWI and fascinated by aviation, you
should certainly consider the opportunity to become an Avi-
onics engineer. EWI provides an excellent Avionics educa-
tion program that prepares you for a future where you can
successfully compete with the best of the best.
Figure 3: Interior of NASA GV.
26 Maxwell 13.3 April 2010
Innovatie in TrammetjeslandEen van de meest onderstreepte features van de nieuwe TU campus is de tramlijn die door de TU-wijk
zal gaan rijden. Prof. Lou van der Sluis vertelt ons in een interview een paar leuke weetjes over de verni-
euwingen op het gebied van bovenleidingloze trams en wat dit te betekenen heeft voor tramlijn 19, die
door het Mekelpark zal lopen.
Auteurs: Ben Allen en Benjamin Gardiner
De aanleg van tramlijn 19 is al aardig op
weg. Merkbaar zelfs, voor iedereen die 18
maart op de TU wijk aanwezig was. Tij-
dens graafwerkzaamheden is een 10kV
kabel geraakt, die dezelfde dag nog ver-
vangen moest worden. Als consequentie
hiervan kwam de TU-wijk een avond lang
zonder elektriciteit te zitten, waardoor de
bezigheden neergelegd moesten worden
en alle gebouwen ontruimd werden. Ook
gaan er geluiden over de campus dat de
faculteit Technische Natuurwetenschap-
pen niet erg blij is met de tram vanwege
storende invloeden
die de bovenleiding
zou veroorzaken.
Op zoek naar de
waarheid gingen
bovenstaande auteurs op bezoek bij pro-
fessor Lou van der Sluis, om te vragen
welke van die geruchten waar zijn en wat
we kunnen verwachten bij de implemen-
tatie van tramlijn 19. In rood de vragen
danwel opmerkingen van bovenstaande
auteurs, in zwart de reactie van professor
van der Sluis.
Wij hadden gehoord dat u iets meer wist
over de draadloze tram die door het Me-
kelpark komt en daar wilden wij graag
iets meer over weten.
Er komt een tram door het Mekelpark,
lijn 19, maar dat heeft een vervelend ne-
veneffect. De bovenleiding van de tram,
daar loopt natuurlijk een stroom door, die
een magneetveld gaat produceren. Dat
magneetveld zou in een aantal gebou-
wen, en met name TNW (Technische Na-
tuurwetenschappen, red.), zeer gevoelige
metingen kunnen
beïnvloeden. Dit
is natuurlijk erg
vervelend, dus wat
we gedaan hebben
is in een klein comité een compensatie-
systeem bedenken. Dat compensatiesy-
steem is bedacht door Prof.dr.ir. P. Kruit
en dat is later naar de markt gewerkt door
een projectgroep waar ik de voorzitter van
was.
“ER KOMT EEN TRAM DOOR HET MEKELPARK, LIJN 19, MAAR DAT HEEFT EEN VERVELEND NEVENEFFECT.”
Je kunt de bovenleiding zo uitvoeren dat
je geen last meer hebt van de magneet-
velden op een afstand van 100 meter.
In de trammetjeswereld is er echter een
ontwikkeling gaande dat trammetjes op
batterijen en accus beginnen te rijden,
en tegenwoordig gaat dat ook met een
supercapaciteit. Dat is het College van
Bestuur ter ore gekomen, en die hebben
toen gevraagd: “Kunnen we niet van die
bovenleiding af?” Dat scheelt kosten, en
de bovenleiding verpest het uitzicht niet.
Dus wordt er ook gekeken naar de moge-
lijkheid van zo’n accutram. Maar je hebt
ook een maatschappij die de tram moet
laten rijden – de HTM – en die heeft een
plan voor het aanschaffen van nieuw ma-
terieel. De oplossing hier moet ook in
dat plan passen. Dus of er over het Me-
kelpark een trammetje zal rijden op ac-
cus of supercapaciteit, dat weet ik niet. Je
moet sowieso wel kijken naar innovatie
in Trammetjesland.
Met het oog op de route van lijn 19 door
Ypenburg, is het nou het idee dat de tram
alleen in het Mekelpark zonder bovenlei-
ding zou rijden, of over het hele traject?
Nou, de actieradius van zo een tram is erg
beperkt, dat is eerder honderden meters
dan kilometers, dus in het uiterste geval
ben je na een of twee kilometers wel klaar.
In Nice hebben ze zo een tram rijden - als
je namelijk een beschermd stadsgezicht
hebt is dat een leuk alternatief. Het voor-
WAT IS EEN SUPERCONDENSATOR?
Supercondensatoren hebben een capaciteit die
vele malen groter is dan die van een conventi-
onele elektrolytische condensator. De techniek
die toegepast wordt stelt men in staat om veel
hogere energiedichtheden te halen. Op moment
van schrijven is de hoogste energiedichtheid die
geproduceerd wordt 30 W·h/kg.
Figuur 1. Een 2.5V 350F supercondensator.
Maxwell 13.3 April 2010 27
deel van zo een tram op accus is dat als de
tram remt, kan je die energie opnemen in
de accus en weer optrekken door gebruik
te maken van deze energie. Per saldo, van
de energiekosten uitgerekend, zou je die
tram wel eens terug kunnen verdienen.
Wij zijn eigenlijk wel nieuwsgierig naar
het compensatiesysteem. Hoe hebben
jullie dat voor elkaar gekregen?
Je hebt wel eens gehoord van een magne-
tische dipool, denk ik. Een magneet is ui-
teraard een magnetische dipool. Maar als
je een door een draadje een stroom stuurt,
in een kringetje, dan krijg je een magne-
tisch veld en dan heb je ook een magne-
tische dipool. En op een afstandje weet
je niet of het een statisch veld is van een
draadje of van een magneet. Die magneti-
sche dipool komt dus doordat dat draadje
een oppervlak heeft. Dus als je je voorstelt
dat de bovenleiding de stroom aanvoert,
gelijkstroom, gaat die door de pantograaf,
door de motor, en via de rails weer terug
naar de voeding. Nu zie je dat daar een
enorme lus in zit - dat is de dipool. En de
vraag is dus “hoe krijg ik die dipool weg?”
Wat we bedacht hebben is om de dipool
weg te halen door de draadjes heel dicht
bij elkaar te leggen.
Je knipt de bovenleiding in stukjes. De
palen van die bovenleiding staan op 40
meter van elkaar en de bovenleiding knip
je door bij elke paal. Als je in gedachten
houdt dat de retourstroom via de rails
gaat en als je nou de voedende kabel langs
de rails legt en je trekt hem bij de paal
op en verbindt hem met de bovenleiding,
dan heb je dus maar een heel klein lusje
gecreeerd en die tram gaat van lusje naar
lusje naar lusje. Op die manier heb je die
dipool enorm verkleind.
Ik vraag me af hoe die supercondensato-
ren danwel accus van energie voorzien
worden, ik neem aan dat ze die uit de bo-
venleiding halen?
Ja ja, dat is zo. Hetzelfde idee wordt ge-
bruikt voor de Rijngouwelijn, dus Gouda
- Leiden - Katwijk, daar willen ze ook
bepaalde stukken geen bovenleiding heb-
ben, daar ben ik ook een beetje bij betrok-
ken. Je zou dus kunnen denken dat je bij
elke halte, zodra de tram stilstaat, hem
even oplaadt, dan heb je helemaal geen
bovenleiding meer nodig. In Nice is het
zo, dat hij in stukjes zonder bovenleiding
rijdt en op sommige stukken wel. Dat is
een mooie oplossing voor het opladen.
Bij de Rijngouwelijn is het toch zo dat Lei-
den aardig dwarsligt? Wordt het een tra-
ject met alleen bovenleiding of maakt de
accutram nog een kans?
Er is nog niets besloten. De provincie
heeft nou een studie bij mij uitbesteed
om te kijken of dat haalbaar is of niet.
De kogel is dus nog niet door de kerk.
Figuur 2. Een tram in Amsterdam
ACCU’S
Een andere mogelijkheid dan de super-
condensator is de welbekende Li-ion
accu. Dezelfde technologie die in je lap-
top zit kan, indien groot genoeg uitge-
voerd, trams en auto’s laten rijden. Be-
kend voorbeeld: De Toyota Prius.
28 Maxwell 13.3 April 2010
The following description of an American breakfast is taken from www.busi-
nessdictonary.com: American breakfast generally includes most or all of the
following: two eggs (fried or poached), sliced bacon or sausages, sliced bread
or toast with jam/jelly/butter, pancakes with syrup, cornfl akes or other ce-
real, coffee/tea, orange/grapefruit juice.
Right. I think that having such a breakfast every morning is probably not
the healthiest thing to do. I have to admit, however, that now and then I do
prepare American pancakes for breakfast. Once, during a conference visit in
the US, I had pancakes and I was immediately hooked. They tasted really
good and since then I occasionally turn the kitchen into a pancake factory.
Ingredients (for 3 to 4 persons) Flour (300 grams)
Buttermilk (0.5 liter)
Four teaspoons of baking powder
Four eggs
Two or three tablespoons of granulated
sugar
Two tablespoons of melted butter
One teaspoon of salt
Cooking with…Rob Remis
American Pancakes
American pancakes (also known as hotcakes, griddlecakes, or fl apjacks) are easy
and fun to make, especially for a leisurely Sunday breakfast or brunch. Before
messing up your kitchen, you might want to set the breakfast table fi rst, and make
sure that everything looks nice. Now the real “cooking” can start. First, combine
all the dry ingredients (salt, sugar, fl our, and baking powder). Then whisk the eggs
and the milk and blend in the melted butter. Finally, mix everything until there
are no lumps left. Cook your pancakes on a hot griddle (a fl at and rimless pan).
You can check the temperature by pouring a little bit of water on the surface of the
griddle. When the water sizzles, the griddle is ready for action. For each pancake,
pour ¼-cup of batter onto the hot surface of the griddle and turn your pancake over
as soon as bubbles start to appear. This is it, basically. There is really nothing to it.
To keep your pancakes warm, place them in a mildly hot oven.
Serve your pancakes with butter and syrup and add your own favorite toppings. I
like to serve my pancakes with bacon and fresh fruit (bananas or strawberries, for
example), but other tasty toppings such as chocolate chips, powdered sugar, jam,
or raisins also go very well with pancakes.
Enjoy!
Maxwell 13.2 January 2010 29
HIER INVOEGEN:PAGINA29MASTERVOLT.PDF
30 Maxwell 13.3 April 2010
TTEthernet
A Powerful Network Solution for All Purposes
Author: Dr. Markus Plankensteiner, Head of Marketing TTTech Computertechnik AG
From Ethernet to TTEthernetOver the last years there has been discussion about how to
adapt Ethernet for new application domains. Its focus has
been on the use of Ethernet for real-time tasks in industrial
environments. More than 20 different approaches are now
struggling for recognition in industrial automation alone.
Today’s Ethernet systems have limits when it comes to
combining them with classical Ethernet networks, de-
vices and services. The scalability of these systems is also
limited and the network solution is tailored for a specifi c
application area. TTEthernet combines the proven deter-
minism, fault-tolerance and real-time properties of the
time-triggered technology
with the fl exibility, dynam-
ics and legacy of “best effort”
of Ethernet and is therefore
suited for all types of appli-
cations.
Considering networking
technologies in general, one
can distinguish between
closed, statically confi gured
embedded networks and open, dynamic networks allowing
free-form communication. Whereas statically confi gured
communication networks enable reliable data transmis-
sion in real time, free-form communication networks per-
form only on a best-effort basis, i.e., there is no guarantee
if and when data messages are transmitted. As statically
confi gured systems usually need to comply with strict safe-
ty are based on dedicated standards, the number of com-
munication nodes is known and not very fl exible. Open
standards, like the TCP/IP/Ethernet stack that drives the
Internet, form the basis of free-form communication and
the number of nodes in systems is arbitrary. TTEthernet
brings together the high fl exibility of free-form systems and
the reliability and speed of statically confi gured systems
(see Figure 1).
TTEthernet Design ObjectivesTTEthernet enables the seamless communication of all
applications by way of Ethernet. Conventional PCs, web
and offi ce devices, multimedia systems, real-time systems
and safety-critical systems are to use the same network. A
single network that is completely compatible with the IEEE
Ethernet 802.3 standards is suited for data transmission
among different applications with various requirements. A
single network solution could thus be used for all appli-
cations in airplanes, ranging from the entertainment pro-
gram, to board supply, electronic navigation and guidance
system, and internet access in passenger seats. Critical ar-
eas are accordingly made fail-safe or fail-operational. Fault
tolerance mechanisms avoid the fault propagation in the
system and prevent potential hackers from unauthorized
access to resources.
TTEthernet is scalable. Networks that connect uncriti-
cal applications shall be able to transmit real-time data in
distributed controls and shall be suited for safety-critical
applications in the future. Existing applications need not
be changed when the network is extended in terms of func-
tionality. Time-critical messages always take precedence
over less important messages in TTEthernet. This does
not affect conventional applications. The temporal behav-
ior of the time-critical messages is predictable and can be
characterized depending on the required quality.
Fig. 1: TTEthernet combines closed-
world and open-world systems on the
basis of IEEE 802.3 Ethernet standards.
Maxwell 13.3 April 2010 31
TTEthernet is used for safety-critical fail-operational ap-
plications. This means that the system remains fully func-
tional even if a failure occurs No matter if a node, a switch
or a network branch is faulty, the network continues safe
communication. This fact accounts for the essential dif-
ference between TTEthernet and other Safe Ethernet sys-
tems. The latter systems, which are sufficient for industrial
applications detect faults in the network and switch the
system to a safe state, e.g. stopping the engine. In order
to secure the availability of the system even if a failure oc-
curs, TTEthernet provides a variety of network services
such as a clock synchronization service, a startup service
and clique detection and recovery services. The behavior
of TTEthernet is precisely predictable and thus formally
verifiable.
TTEthernet System PropertiesTTEthernet has time-triggered services that enable time
triggered communication over Ethernet. These time-trig-
gered services establish and maintain a global time, which
is realized by the close synchronization of local clocks of
the devices. The global time forms the basis for system
properties such as temporal partitioning, precise diagnosis,
efficient resource utilization, or composability.
Temporal Partitioning: The global time can be used as
a powerful isolation mechanism when devices become
faulty; we say that the global time operates as a “temporal
firewall”. In case of failure it is not possible for a faulty ap-
plication to untimely access the network. Depending on
the location of the failure, either the communication con-
troller itself or the switch will block faulty transmission
attempts. Failures of the switch can be masked by powerful
end-to-end arguments such as CRCs or by high-integrity
designs.
Efficient Resource Utilization: The global time contrib-
utes to efficient resource utilization in several ways. Time-
triggered communication allows minimizing the memory
buffers in network devices as the time-triggered commu-
nication schedule is free of conflicts. Hence, switches do
not have to be prepared for bursts of messages that have to
be delivered over the same physical link. A minimal time-
triggered switch design could even multiplex media access
logic such as reception or transmission logic. A second
way of effective resource utilization is buffer memory in
the nodes, which can be minimized as the sensor values
can be acquired according to the global time, immediately
before sending the message. Finally, a third way of effective
resource utilization is power management in which energy
can be seen, and saved, analogously to memory.
Precise Diagnosis: A global time stamping service simpli-
fies the process of reconstruction of a chain of distributed
events. On the other hand, the synchronous capturing of
sensor values allows building snapshots of the state of the
overall systems.
Composability: The global time allows the specification of
devices not only in the value domain, but also in the tem-
poral domain. This means that already during the design
process of devices, the access pattern to the communica-
tion network can be defined. The devices can then be de-
veloped in parallel activities. Upon integration of the indi-
vidual devices, it is guaranteed that prior services are stable
and that the individual devices
operate as a coordinated whole.
Dataflow Options in TTEthernetTTEthernet specifies services
that enable time-triggered com-
munication on top of Ethernet.
The time-triggered services can
be viewed parallel to the usual
OSI layers: a communication controller that implements
these services is able to synchronize with other communi-
cation controllers and switches in the system. It can then
send messages at points in time derived from this system-
wide synchronization. These messages are then called
time-triggered messages.
As TTEthernet supports communication among applica-
tions with various real-time and safety requirements over
a network, three different traffic types are provided: time-
triggered (TT) traffic, rate-constrained (RC) traffic, and
best-effort (BE) traffic. If required, the corresponding traffic
type of a message can be identified based on a message’s
Ethernet Destination address. The relation of the TTEth-
ernet traffic types to existing standards is depicted in Figure
2.
Messages from higher layer protocols, like IP or UDP, can
be “made” time-triggered without modifications of the
messages’ contents itself. The TTEthernet protocol over-
head is transmitted in dedicated messages termed protocol
control frames, which are used to establish system-wide
synchronization. In short, TTEthernet is only concerned
Fig. 2: Relation of TTEthernet to
existing communication standards
32 Maxwell 13.3 April 2010
with “when” a data message is sent, not with specific con-
tents within in a message.
TT messages are used for time-triggered applications. All
TT messages are sent over the network at predefined times
and take precedence over all other traffic types. TT mes-
sages are optimally suited for communication in distrib-
uted real-time systems. TT messages are typically used for
brake-by-wire and steer-by-wire systems that close rapid
control loops over the network. TT messages allow design-
ing and testing strictly deterministic distributed systems,
where the behavior of all system components can be speci-
fied, analyzed and tested with sub-micro second precision.
RC messages are used for applications with less stringent
determinism and real-time requirements than strictly
time-triggered applications. RC messages guarantee that
bandwidth is predefined for each application, and delays
and temporal deviations have defined limits. RC messages
are used for safety-critical automotive and aerospace appli-
cations that depend on highly reliable communication and
have moderate temporal quality requirements. Typically,
RC messages are also used for multimedia systems.
In contrast to TT messages, RC messages are not sent with
respect to a system-wide synchronized time base. Hence,
different communication controllers may send RC mes-
sages at the same point in time to the same receiver. As a
consequence, the RC messages may queue up in the net-
work switches, leading to increased transmission jitter. As
the transmission rate of the RC messages is bound a priori
and controlled in the network switches, an upper bound on
the transmission jitter can be calculated off-line and mes-
sage loss is prevented.
BE messages follow a method that is well-known in clas-
sical Ethernet networks. There is no guarantee whether
and when these messages can be transmitted, what delays
occur and if BE messages arrive at the recipient. BE mes-
sages use the remaining bandwidth of the network and
have less priority than TT and RC messages. Typical user
of BE messages are web services. All legacy Ethernet traffic
(e.g. internet protocols) without any QoS requirement can
be mapped to this service class. TTEthernet implements
strong partitioning between non-critical BE traffic and all
other service classes (see Figure 3).
TTEthernet as Transparent Synchronization Pro-tocolTTEthernet is a transparent synchronization protocol, i.e.,
it is able to co-exist with other traffic, potentially legacy
traffic, on the same physical communication network. For
reasons of fault tolerance a multitude of devices can be con-
figured to generate synchronization messages. The devices
generating the synchronization messages may be distribut-
ed with a high number of intermediate devices in between
each other.
Fig. 4: A test setup for TTEthernet
Fig. 3: TTEthernet includes TT, RC and BE messages.
Maxwell 13.3 April 2010 33
TTEthernet defi nes basic building blocks that allow the
transparent integration of the time-triggered services on
top of message-based communication infrastructures such
as standard Ethernet. For this, TTEthernet defi nes a novel
application of the transparent clock mechanism that ena-
bles the concept of the permanence point in time, which
allows re-establishing the send order of messages in a re-
ceiver:
Application of transparent clock mechanism: all devices
in the distributed computer network that impose a
dynamic delay on the transmission, reception, or relay of
a synchronization message add this dynamic delay into a
dedicated fi eld in the synchronization messages used for
the synchronization protocol.
Novel precise calculation of the permanence point in
time: the application of transparent clock mechanism
allows a precise re-establishment of the temporal order
of synchronization messages. In a fi rst step the worst
case delay is calculated off-line. In a second step, each
synchronization message is delayed for ”worst case delay
minus dynamic delay” upon reception of the synchroniza-
tion message, where the dynamic delay is the delay added
to the synchronization message, as the synchronization
message fl ows through the communication channel. This
point after the reception point in time will be called the
permanence point in time.
For fault-tolerant algorithms in general, and fault-tolerant
synchronization algorithms in particular, the message send
order is of highest importance. The re-establishment of
the send order of synchronization messages is required for
any fault-masking synchronization protocol that ensures
synchronization of local clocks in a distributed computer
network.
Safety and Fault ToleranceA high level of safety is provided by the time-triggered
method of TTEthernet, which detects failures and irreg-
ularities in the network and certain systems. Additional
measures need to be taken to achieve maximum safety,
availability and fault tolerance.
Contact: TTTech Computertechnik AG
Schoenbrunner Strasse 7
A-1040 Vienna, Austria
Tel.: +43 1 585 34 34-0
Fax: +43 1 585 34 34-90
E-mail: [email protected]
Web: www.tttech.com
TTEthernet networks can be set up with multiple redun-
dant end systems, switches and segments. Thus the system
will remain in operation even if faults occur. Redundant
network paths are always used in fault-tolerant TTEther-
net systems so that the failure of a single system or mes-
sages can be tolerated without affecting the application. If
multiple redundancy is implemented, multiple faults can
be tolerated. It is important that the entire system remains
in operation without interrupts under the same temporal
conditions as defi ned before.
TTEhernet allows the integration of guardians in switches
and end systems. Guardians check if the communication
on the network works in compliance with the predefi ned
parameters. If faulty systems block network segments, the
guardian disconnects the network segment or port. Multi-
ple redundant guardians can be implemented to meet the
highest safety requirements).
ConclusionTTEthernet enables time-triggered communication over
Ethernet networks in all application areas. The network
provides all necessary mechanisms for applications as di-
verse as classical web services and time-critical and safety-
critical control system in airplanes. Existing networks can
be extended step by step using TTEthernet-capable switch-
es and end systems without the need to change existing
applications and end systems. Reducing network solutions
to established and recognized Ethernet standards opens up
saving potentials that secure major advantages in competi-
tive markets.
Fig. 5: A network card as used by TTTech for testing.
34 Maxwell 13.3 April 2010
Joost may know itHoe werkt een plasmabol?
Author: Benjamin Gardiner
De plasmabol is een voorwerp dat in het kader valt van fascine-
rende objecten, net als de lavalamp. Kenmerkend aan deze ob-
jecten is dat je er uren naar kan staren met de vraag hoe ze nou
werken. Vandaag zal ik jullie vertellen hoe één van deze objecten
werkt, namelijk de plasmabol.
Je hebt vast gezien hoe in een plasmabol alle bliksemflitsen naar
de buitenkant van de bol gaan en hoe je dit effect kan verstoren
door je hand op het glas te leggen. Als je dit doet gaan de blik-
semflitsen naar je hand toe, maar waarom?
Een plasmabol is opgebouwd uit drie verschillende onderdelen.
De eerste is de elektrode in het midden van de plasmabol en
dient voor het opbouwen van het potentiaalverschil. Het tweede
onderdeel is het medium waar de bliksemflitsen zich doorheen
bewegen, dit is meestal een edelgas maar een gas als stikstof kan
ook. De reden waarom lucht niet geschikt is, komt doordat lucht
geen duidelijke vonken maakt en het potentiaalverschil een stuk
hoger moet zijn dan bij edelgassen. Het derde deel van de plas-
mabol is de glazen bol om de plasmabol. Deze dient simpelweg
voor het vasthouden van het gas.
Wat is een plasma? Plasma wordt ook wel gezien als de vierde ag-
gregatietoestand. Plasma ontstaat door atomen op te warmen tot
heel hoge temperaturen. Daardoor beginnen de atomen met gro-
te snelheden te bewegen zodat, telkens als ze botsen, de elektro-
nen weggerukt worden uit de atomen. Zodoende bestaat plasma
uit positief geladen ionen en de elektronen die van de atomen
zijn losgekomen. Dit kan ook gerealiseerd worden door een hoog
potentiaalverschil aan te brengen waardoor de elektronen van de
kernen worden los getrokken.
Wat is nou de reden dat die bliksemflitsen naar je hand toe gaan
als je deze op het glas legt. Dit komt doordat de ionen die zijn
ontstaan bij het ioniseren van het gas nieuwe elektronen zoeken.
Door jouw hand op de glazenbol te leggen creëer je een aardpunt
waar nieuwe elektronen vandaan kunnen komen. Zo ontstaat er
door je lichaam naar aarde een stroom, deze voel je niet omdat
hij zeer klein is. De ionen zoeken nieuwe elektronen en worden
daardoor aangetrokken naar het aardpunt wat in dit geval jouw
hand is, dit zie je vervolgens als bliksemflitsen die in de richting
van jouw hand bewegen.
Plasmabollen
Maxwell 13.2 January 2010 35
HIER INVOEGEN:PAGINA35NAVIGON+PINEWOOD.PDF
36 Maxwell 13.3 April 2010
The Nuon Solar Team and their famous car the Nuna had a great tradition of winning the World Solar
Challenge in Australia. Four times in a row the car was faster than all the other solar cars. This year,
however, it was different. Some headlines claimed we lost first place, others said we won second. Be-
cause of all the hard work our team put in the car, I would agree with the second one. We had a great
fight with the Michigan solar car and managed to stay in second position. This year’s winning car, from
Japan, was way ahead of us from the first day on. So, what went wrong with the legendary Nuna? Why
was the fifth victory out of our reach?
Before the raceThree weeks before the race in Darwin, Australia, the car
was getting into shape and testing went well. We could fi-
nally start to change our team from a group of engineers to
a well-oiled racing team. At the point where none of us ex-
pected it, it all went terribly wrong. The rear tire blew and
Nuna became unstable. The driver could no longer control
the three wheeled vehicle because the most import one of the
three was no longer there.
Nuna went off-road with a speed of approximately 100km/h
which resulted in a spectacular crash. Large parts of the so-
lar panels were damaged, most of the suspension was gone
and, of course, a lot of bodywork was ripped away. Luckily
most of the structural parts were still intact and probably the
most imported part, the driver, could get out without a single
scratch. After an analysis of what had happened we pulled
the team back together and started the rebuild. With a lot of
support from our sponsors we managed to get the car run-
ning again within two weeks. We could finally start testing
again but we did lose a valuable 2 weeks of race preparations.
Four days of troubleDuring the race the electronics of our precious lady Nuna
performed a bit disappointing. A lot of systems that we used
for monitoring the car stopped working, or only worked a
part of the time. The most important issue was a failing
MPPT (Maximum Power Point Tracker). Even when we re-
placed the tracker with a fresh one it would stop working
within an hour. And, most surprisingly, the old one worked
fine in the test environment. It took us four days to find the
problem. What was the cause of this stupid problem I will
discuss later. First I will go into the details of the tracker.
IV CurveThe first question you might have is why we would need an
MPPT between the solar panel and the battery of the car.
Well, as the name suggest, it is to get a maximum amount
of power from the sun. This all depends on the voltage curve
of solar cells.
When a photon hits a solar cell an electron is kicked out of
the semiconductor material. The intensity of the light deter-
mines the flow of electrons. If a constant amount of photons
per second hit the material, a constant amount of electrons
will be kicked out of the cell. This behavior can best be mod-
eled whit a current source.
To get the maximum amount of power out of a current source
we should keep the voltage as high as possible. However if we
do this with a solar cell we will suffer from recombination
Figure 1: The Nuna5 after crashing.
Nuna5:What went wrong?Author: Rico van Dongen
Maxwell 13.3 April 2010 37
and the cell starts dissipating most of the energy on its own.
With some cells you can even see this effect when the cell is
left in the sun without a load. Suddenly small red lights ap-
pear on these cells. Also hotspots can appear that sometimes
become so hot that they burn trough the cell (Figures 2&3).
In figure 4 an IV curve (blue line) of an arbitrary silicon solar
panel is given. The red line in this figure shows the multi-
plication of voltage and current, the PV curve. It is clearly
shown that the maximum amount of power can be drawn
from this panel by adjusting the output voltage to the point
where it starts going down. This lower voltage prevents the
electrons from recombining with their holes and makes sure
that all the power goes to the battery.
MPPTThe maximum power point (MPP) of a solar cell is constantly
shifting due to changing weather conditions. When the sun
intensity is low the MPP shifts to a lower voltage. Of course
a battery can only be charged when the voltage is higher than
its own voltage so there need to be some sort of conversion.
That is why the tracker is basically an adjustable boost con-
verter. A special tracking algorithm tests certain point on the
IV curve and makes sure the converters input voltage is near
the MPP.T These testing points are the yellow dots in figure
4. When the intensity of the sun changes the MPPT has to be
adjusted to the new intensity. Compared to the tracking algo-
rithm these changes are slow so by simply trying different points
the converter goes back to the MPP. Some trackers are also able
to detect the slope of the PV curve and thereby knows if it has to
increase or decrease its input voltage.
Failing connectionsSo, now we know what the tracker should do. But why did it
stop working during the race? Probably the cause was in the
mounting of the board. One support had some minor fab-
rication errors that caused some tension on the PCB of the
tracker. After a while the board would heat up and expand a
bit. This expansion caused some extra tension that resulted
in a broken contact. This stupid problem caused us about
one quarter of the total solar input power. This means that
we came in second with only three quarters of the energy.
Quite an achievement if you ask me.
Better luck next timeOf course we all learned a lot during the design and build
phase of the Nuna project, but we probably learned the most
during the 5 days of intense racing. I am sure that the next
team will put some special care in the supports of all the
PCB’s and improve the design even more. Then, the Nuna6
will be able to win the race again and maybe even break
the world record. This record is still in the hands of Nuna3
which could drive up to 103kph average thanks to the more
flexible regulations.
To make sure the new team knows all the details about
Nuna we now started a special course. Under the cover of
the course ‘sustainable mobility and transport’ (AE4-T39),
we now go through the complete design. If you would like a
taste of the Nuna race spirit and know more about the design
of Nuna5 feel free to follow this course. But of course, if you
really want to know how amazing it feels to design, build and
race a solar car you will have to join the Nuna6 team.
Figure 2: Thermal image - normal. Figure 3: Thermal image - hotspot.
Figure 4: IV-curve.
38 Maxwell 13.3 April 2010
Everybody who has seen a James Bond
movie, or anything involving spying of
any sort, has heard of the concept of a
bug. A small electronic device is placed
in a person of interest’s room, and the
agents in the other room can listen in
on what’s going on. Evidence is gathered,
and the story can move on.
Of course, in reality things are a little dif-
ferent. The microscopic devices used in
Hollywood are slightly more advanced
than what we can achieve on our bread-
boards. Unless you have a very steady
hand or access to an smd soldering oven
you’re not going to be able to replicate the
miniature devices seen in the movies.
Legal concernsThere are a few less frivolous concerns
when it comes to radio circuits. Radio
frequency transmissions are strictly regu-
lated and simply transmitting on AM/FM
bands is illegal - something that has shut
down many decent pirate radio station
in the past. The reason for this is sim-
ple: you need a license to use a certain
frequency - and these licenses are sold at
auction. Because commercial radio sta-
tions are also interested in acquiring a
license to transmit, these become incre-
dibly expensive. As such, transmitting on
AM band is illegal in the Netherlands.
Luckily, there is a provision in the law for
low-power FM devices, for instance to lis-
ten to your iPod in the car without having
to fit a new stereo system.
Please note that the power allowed for
these devices is severely limited. The cir-
cuit we have featured has a range of about
20-30 metres, but more powerful devices
step over the low-power boundaries and
as such are not permitted.
Listening inBefore we can even start to think about
transmissions, we need something to
transmit. In our case, we’re looking to
transmit audio. In figure 1 we have a
simple one-transistor amplifier that takes
Circuit BodgingFM bug transmitter
Many people have their interest in electronics start with radio. As much as radio is taken for granted by
most people, the idea of sending information through the through the air at the speed of light remains
enticing to anyone who thinks about it for longer than a few minutes. As such, we submit to you this
simple yet effective FM transmitter.
Author: Ben Allen
its input from a condensor microphone.
A condensor microphone (figure 2) is a
special type of microphone in which the
diapraghm (the moving part of the mi-
crophone) acts as one of the plates of a
capacitor. As the diaphragm moves, the
capcitance changes.
Figure 1. The condenser microphone
and surrounding circuitry.
Figure 2. A number of electret condenser
microphones.
Maxwell 13.3 April 2010 39
Part list:
R1: 22kR2: 1M R3: 10kR4: 47kR5: 470
C1: 223pFC2: 100nFC3: 102pFC4: 5.6pFC5: 27pFC6: 6-45pF trimmer
L1: 1μH
ANT1: simple antenna
T1, T2: BC547
Using a microphoneThe circuitry in figure 1 is fairly self-
explanatory. One thing to take into ac-
count is that the condenser mic needs a
voltage to be applied across the device for
it to operate. Unlike dynamic micropho-
nes, condensers don’t create an output by
themselves. We are, in essence, measu-
ring and amplifying the change in the de-
vice’s behaviour as opposed to measuring
an output from the device itself.
Turning it into radioThe interesting stuff happens in figure 4.
First of all, note that the input does not
have a DC filtering capacitor. This is be-
cause the microphone amplifier circuit
already features this capcitor, C2. If you
intend to use the circuit to broadcast so-
mething other than the output from the
microphone you should add the 100nF
capacitor in series with the input.
C6 and L1 provide an oscillator, the fre-
quency of which can be adjusted with C6.
When tuning the circuit, tune your radio Figure 4. The RF modulator.
to 90MHz and attempt to recieve a signal
whilst adjusting C6.
Because the capcitance of T2 is in the
same order of magnitude as C6, the idea
is that as the transistor amplifies the in-
put, this capcitance influences the oscil-
lator, causing the frequency to change -
Frequency Modulation.
This is presented, once again through a
DC filtering cap, to the antenna, which
broadcasts the signal.
Power supplyThe power requirement of the circuit is
very simple, as shown in figure 3. A 3-5V
supply is enough, so a series connection
of 3 AAA cells can power the circuit easi-
ly. The large 470μF capacitor C7 provides
smoothing in the case of external power
supplies.
Final NotesMaking an FM transmitter out of discrete
parts is notoriously unreliable. A simpler,
but less interesting way is to use ICs to
provide modulation, as all they require is
an input signal and an antenna, and in
some cases a crystal for the oscillator. If
your goal is reliability as opposed to expe-
rimentation, an IC such as the MC2833
or the MAX2606 might be more to your
liking.
Figure 3. Power supply
40 Maxwell 13.3 April 2010
Via via werd ik gevraagd of ik op een elektrische
scooter namens de gemeente Delft wilde deelne-
men aan de Road to Copenhagen. Als vierdejaars
student Electrical Engineering en Bestuurder van
de Electrotechnische Vereeniging is dit een vraag
waarop maar één antwoord mogelijk is: ja! Niet
alleen is het een geweldige uitdaging, ook is het
een buitenkansje om de Elektrotechniek weer
eens voor het voetlicht te halen.
Toen ik mij meer ging informeren over de actie,
las ik dat een groep van wel honderd jongeren op
scooters naar Kopenhagen zou gaan rijden, om
daar tijdens de klimaatconferentie afgelopen de-
cember een manifest aan de wereldleiders aan te
bieden.
Aanvankelijk was ik daarom bang dat ik met een
stel hippies naar Kopenhagen zou gaan. De groep
was echter erg divers. Zo waren er studenten be-
drijfs- en bestuurskunde, internationale betrek-
king en duurzame ontwikkelingen. Ik was een van
de weinige technische studenten. Ondanks deze
ogenschijnlijke cultuurbarrière zijn we al snel een
hechte gezellige groep geworden. Het hoofddoel
van de tocht was, naast het promoten van duurzaam vervoer, het aan-
bieden van een manifest aan de wereldleiders die op dat moment aan de
klimaattop deelnamen.
De scooter waar ik op reed werd gesponsord door Delft en was onder
handen genomen door het Delftse bedrijf Epyon. Daarom kon deze in
een half uurtje volgeladen worden, in tegenstelling tot de andere negen-
tig scooters. Voor de gemeente Delft was deze tocht een goede manier
om hun betrokkenheid bij duurzame ontwikkelingen te tonen. Per dag
werd er vier uur gereden. Geheel conform de tijd van het jaar was het
rijden erg koud. Gewapend met thermo-ondergoed, een
fl eecetrui en een dikke winterjas reed ik in vijf dagen de
eerste etappe, van Den Bosch naar Osnabrück.
Wanneer we niet op de scooter zaten, werden we ver-
maakt met lezingen, workshops of excursies. Deze dien-
den als inspiratie voor het manifest dat we aan het einde
van de tocht in Kopenhagen aan de wereldleiders hebben
aangeboden. In dit manifest staan twaalf tips om een
duurzamere samenleving te verwezenlijken. Zo kwam
mijn groepje met het idee om het voor consumenten
duidelijker te maken dat zuinigere producten op de lange
duur goedkoper zijn. In de supermarkt staat verplicht op
Samen met honderd andere studenten reed de Com-
missaris Onderwijs van de ETV, Frank Teunisse, de
‘Road to Copenhagen’ om tijdens de klimaattop
aandacht te vragen voor duurzaam vervoer.
Per elektrische scooter naar KopenhagenAuteur: Frank Teunisse
Maxwell 13.3 April 2010 41
Naast al deze serieuze activitei-
ten werden we goed verzorgt;
iedere dag, soms zelfs al in de
lunch, stond er een heerlijke
warme maaltijd op ons te wach-
ten, ter compensatie voor de
koude ontberingen van die dag.
Alsof dat niet genoeg was, werd
de dag afgesloten met een goed
feest. Echter, na een korte nacht
in een jeugdherberg moesten we
voor acht uur al weer klaar staan
voor de volgende dag.
Het meest indrukwekkend vond
ik zonder twijfel het meerijden
in een Tesla Roadster. De Tesla
is een elektrische sportauto, ver-
noemt naar uitvinder en Elec-
trotechnisch ingenieur Nikola
Tesla (1856 – 1943). Met recht heeft deze wagen de titel ‘sportauto’, aan-
gezien de ‘0 tot 100km per uur’ al in 3.7 seconde gehaald kan worden. En
dat deed hij dan ook. Het voelde als in een achtbaan, ik werd helemaal in
mijn stoel gedrukt. Het piepen van de banden was daarbij het enige wat ik
hoorde, want de motor produceert slechts een miniem gezoem.
Al met al ben ik zelf niet veel “groener” geworden van deze ervaring. Ik
ben me wel veel meer van bewust van het klimaatprobleem en ik geloof nu
wel dat dat ook echt een probleem is. Momenteel wachten consumenten
en het bedrijfsleven op elkaar om te investeren in bijvoorbeeld elektrisch
vervoer. Aan zet is, wat mij betreft, dan ook de overheid. Zij kan zowel de
burgers als het bedrijfsleven stimuleren te investeren in duurzame pro-
ducten. Deze reis heeft mij in ieder geval kunnen overtuigen dat elektrisch
vervoer een volwaardig alternatief voor vervoer op brandstof is. Nu is het
wachten op de grote massa.
alle producten hoeveel calorieën een portie bevat.
Wanneer er op een lamp verplicht staat aangege-
ven hoeveel deze per jaar kost, zijn mensen veel
eerder geneigd een duurdere, maar in gebruik
goedkopere, LED-lamp te kopen.
De meest interessante lezing was die van Mi-
chael Braungart, Hoogleraar Cradle-to-Cradle
aan de Erasmus Universiteit Rotterdam. Zijn
idee is dat energie bespáren alleen tot gevolg
heeft dat men minder slecht is voor het milieu.
Waar men vanuit zou moeten gaan, is dat je iets
positiefs aan het milieu toevoegt, in plaats van zo
min mogelijk negatiefs. Zo heeft Braungart mee-
gewerkt aan een duurzame wolkenkrabber in de
woestijn, die meer energie en water opbrengt dan
de inwoners zelf gebruiken.
42 Maxwell 13.3 April 2010
Conceptual ‘CTRUS’ football gets loaded with sensors, don’t need no pumpWe’ve heard of soccer
balls that play a tune
when kicked, sure,
and we’re pumped
to see the World
Cup in 3D, but it’s
not often that someone
comes up with a serious
technological makeover for
the sport that’s nearly as old as
life itself. CTRUS, however, is just
that - a theoretical revolution in soccer
that begins with the all-important ball. To start with, a reinforced elastic
structure means that CTRUS doesn’t require any air. (So long, pump.)
Next, GPS and RFID chips keep track of the ball’s position at all times,
and tell it to light up in different colors when it scores a goal or is ac-
complice to a nefarious violation. (Farewell, referee.) Last but not least,
the sphere itself will report back with accelerometers that measure the
ball’s kick force and travel speed, and a camera that could (with magical
software stabilization, of course) actually fi lm action from the ball’s own
POV. Sadly, the ball is just a concept from an undercover
marketing agency, but since we’re dreaming, we
urge its creators to add a second camera. Just
imagine just how immersive it would be to
have your face booted in at 130km/h in glo-
rious 3D. Or, just peek the concept videos
after the break.
Source: www.gizmodo.com
Tiny Generators Charge Up From Random Vibrations In the AirWhat’s an easier green power source to acquire than
sun or wind? Random ambient vibrations, that’s what.
And that’s exactly how some new generators juice up.
The devices, created by researchers at the University
of Michigan, aren’t going to be powering your cars any-
time soon, but they’ll be able to gather enough energy
to power watches, pacemaker or wireless sensors.
The vibrations they pick up are from things like traf-
fi c on bridges, machinery operating in factories or you
swinging your arms around.
The researchers have built three prototypes and a
fourth is forthcoming. In two of the generators, the
energy conversion is performed through electromag-
netic induction, in which a coil is subjected to a vary-
ing magnetic fi eld. This is a process similar to how
large-scale generators in big power plants operate.
The latest and smallest device, which
measures one cubic centimeter, uses
a piezoelectric material, which
is a type of material that
produces charge when it is
stressed. This version has
applications in infrastruc-
ture health monitoring.
The generators could one
day power bridge sensors
that would warn inspec-
tors of cracks or corrosion
before human eyes could
discern problems.
Maxwell 13.3 April 2010 43
Auteur: Technolution B.V.
Gedreven door ontwikkelingen in de 3D-
gamingmarkt zijn zeer snelle grafische
processors ontwikkeld. Bij de eerste gra-
fische toepassingen rekende de CPU van
de pc pixel voor pixel het weer te geven
beeld uit. Daarna werd pixel voor pixel
het beeld naar de videokaart gestuurd.
Vervolgens gaf de videokaart slechts de
pixels weer. Dit is een zeer rekeninten-
sieve klus, zeker als dit bijvoorbeeld in
een 3D-game bij 30 frames per seconde
moet gebeuren met een resolutie van
1920 x 1200. In de jaren negentig kwa-
men steeds krachtigere videokaarten
voor de consument op de markt en kreeg
de videokaart zijn eigen processor, de
GPU. In plaats van pixel voor pixel de
videokaart aan te sturen worden objec-
ten en instellingen aangeleverd, zodat
deze aparte processor zelfstandig werk
kan verrichten naast de CPU.
StandaardisatieIn de 3D-gaming, een consumenten-
markt, worden zeer grote volumes van
deze processors verkocht. Omdat de
producenten van de GPU’s en de games
verschillende bedrijven zijn, ontstond de
noodzaak tot standaardisatie van de in-
terfaces (API’s). De belangrijkste 3Din-
terfacestandaarden zijn DirectX/Di-
rect3D en OpenGL. In de praktijk komt
het erop neer dat de GPU-producenten
strikt de gespecificeerde functionaliteit
van deze 3D-interfaces (API’s) volgen.
ATI (nu AMD) en NVIDIA zijn bekende
producenten van GPU’s. De kern van
3D-spellen is dat een virtuele 3D-wereld
op het scherm wordt afgebeeld vanuit
het perspectief van een (virtuele) speler,
met een snelheid van bijvoorbeeld 30
frames per seconde. Dit wordt gereali-
seerd door een hele reeks bewerkingen
in een vaste volgorde uit te voeren. Dit
proces heet een rendering pipeline.
GeschiedenisIn de jaren negentig kwamen steeds
krachtigere videokaarten voor de consu-
ment op de markt. En een steeds groter
deel van de rendering pipeline werd in
hardware uitgevoerd. De bewerkings-
stappen in de rendering pipeline zijn de
zogenaamde shaders. Bijvoorbeeld een
vertex shader voegt bepaalde 3D-effec-
ten toe aan objecten, een geometries-
hader genereert nieuwe objecten vanuit
al gedefinieerde objecten en een pixels-
hader berekent de kleur van een pixel.
Een belangrijk nadeel was dat met het
vastleggen van de specifieke shaders in
hardware ook de rekenkracht per shader
vast lag. Dit terwijl verschillende spellen
verschillende eisen stelden aan de verde-
ling van de rekenkracht tussen de ver-
schillende shaders. Relatief recent is er
als onderdeel van de DirectX-standaard
een Unified Shader Model gespecifi-
ceerd. Het betreft een processormodel
dat uitgerust is met een instructieset
die geschikt is voor de functionaliteit
van alle verschillende typen shaders.
Het voordeel van dit concept is dat per
stap van de rendering pipeline de volle-
dige rekenkracht aangewend kan worden
en programmeerbaar is. Met dit nieuwe
type grafische processor is ook de eerste
praktische mogelijkheid ontstaan om de
GPU voor ander rekenwerk in te zetten.
Rekenkracht moeilijk te temmenDe hedendaagse GPU beschikt over een
enorme rekenkracht. Hij kan tot wel 30
keer meer floating pointberekeningen
verwerken dan een reguliere CPU (zie
figuur). Dat klinkt geweldig: zoveel meer
rekenkracht in een processor voor glo-
baal dezelfde prijs. Dat wil iedereen wel.
Maar als iets te mooi klinkt om waar te
zijn, dan is het dat vaak ook. Aan een ‘al-
ternatief ’ gebruik van een GPU, anders
dan als grafische kaart in een pc, kleven
de nodige mitsen en maren. Het reken-
kundige probleem moet goed passen op
de architectuur van een GPU, evenals
de floating pointprecisie en het is
GPU: brute kracht inzetbaar voor meer dan alleen beelden?
De GPU (Graphics Processing Unit) in een moderne pc is een waar
rekenmonster. Bij het produceren van een vloeiende stroom beel-
den voert de GPU meer berekeningen perseconde uit dan de pro-
cessor (CPU). Maar in de praktijk blijkt het niet eenvoudig GPU’s
intezetten voor andere toepassingen.
44 Maxwell 13.3 April 2010
praktisch niet mogelijk om een GPU te
combineren met een andere hardwarear-
chitectuur dan die van een pc.
Dieper in de techniek van de GPUIs het verhaal dan klaar met de conclu-
sie dat de GPU elk jaar betere beelden
maakt, maar een plaatjesmachine is en
blijft? Zo zwart/wit liggen de zaken niet.
In bepaalde gevallen en onder strikte
voorwaarden zou een GPU als alterna-
tief kunnen dienen voor ander zwaar
rekenwerk dan alleen grafisch werk.
Daarvoor moeten we iets dieper in de
techniek duiken. Een GPU is een paral-
lelle processor: hij voert zeer veel paral-
lelle bewerkingen uit. Dat is bijvoorbeeld
te zien aan de brede geheugenbussen, tot
wel 256 bits. Ter vergelijking: de CPU
is nog niet zo lang geleden van 32 naar
64 bits gegaan. Een GPU werkt dus
met veel meer data tegelijk. Maar het
ontwerp is bedoeld voor eenrichtings-
verkeer: er komt een brok data binnen
met elementaire beeldinformatie, voor
elke pixel berekent hij het een en ander,
vervolgens stuurt hij dat er aan de an-
dere kant uit richting het beeldscherm.
Een GPU is slecht in keuzes maken,
zoals in het doorlopen van typische “if
..., then ..., else ...”- softwareconstruc-
ties. Daar wordt hij traag van. Een GPU
is juist gebouwd om voor alle parallelle
paden hetzelfde te doen. Dat heet sin-
gle instruction, multiple data (SIMD).
Er is een instructiedecoder en een hele
groep executie-units. Zodra er een keuze
komt, kan die maar voor een executie-
unit gelden en kunnen de anderen niets
effectiefs doen. Een gewone processor is
veel beter in het oplossen van if-then-
else-problemen. Die kan slim door een
reeks van voorwaarden en afhankelijk-
heden heen slalommen en zich veel re-
kenwerk besparen, als een skiër die in
de afdaling continu kiest en stuurt om
de snelste weg te vinden. Een GPU is
als een roeiboot met bijvoorbeeld acht
roeiers die tegelijkertijd samen een doel
proberen te bereiken. Als elke roeier een
ander parcours zou moeten varen, dan
kan dat alleen maar door acht parcour-
sen achterelkaar te varen met telkens
één roeier aan het werk. Een GPU is
dus goed in het parallel verwerken van
rekenkundige bewerkingen volgens vaste
formules.
Berekeningen aansturen met APIAls een GPU gebruikt gaat worden,
moet dat gebeuren met speciale instruc-
ties, die bij elk merk GPU kunnen ver-
schillen. Hierbij zijn de standaard API’s
niet bruikbaar. Het is dus hard nodig om
toegang tot de GPU’s te standaardiseren,
zodat de functies die moeten worden uit-
gevoerd op een redelijk abstract niveau
kunnen worden aangeboden. Op dit mo-
ment zijn de belangrijkste twee: CUDA
van NVIDIA en de merkonafhankelijke
variant OpenCL (Open Computing Lan-
guage). OpenCL is initieel gedefinieerd
door Apple Inc., maar is door hen via
de Kronos Groep tot een open standaard
gemaakt. Door deze API’s te gebruiken
kunnen de ingewikkelde details van een
GPU worden beheerst en kan de pro-
grammeur zich richten op het onder-
havige probleem. Ook zullen door stan-
daardisatieprogramma’s (bv. Matlab) er
bibliotheken ter beschikking komen met
GPU-ondersteuning voor veel voorko-
mende problemen, zodat niet iedereen
opnieuw het wiel hoeft uit te vinden.
Rekenkracht GPU anders aan-wendenDoor de genoemde API’s is, zoals eerder
aangegeven, het mogelijk om goed pas-
sende rekenkundige problemen, zonder
uitvoer naar een beeldscherm, met suc-
ces sneller op een GPU uit te voeren
dan op een CPU. Gedacht moet worden
aan o.a. matrixberekeningen en Fourier-
transformaties, zoals die bijvoorbeeld
voorkomen in ruimtelijke problemen als
3Dsimulaties van stromingen of elektro-
magnetische velden. Dit zijn verschijn-
selen die zich laten analyseren door de
wereld in kleine vakjes of kubusjes te
verdelen om per vakje een natuurkun-
dige vergelijking op te lossen (ook wel de
eindige elementenmethode genoemd).
Het oplossen van deze vergelijkingen
vergt heel veel ‘simpele’ rekenstappen
op een grote dataset, maar ook andere
grootschalige berekeningen (vooral floa-
ting point) en beeldgeneratie voor de
professionele en wetenschappelijke ge-
bieden. Voor dit soort toepassingen heeft
NVIDIA een professionele GPU ontwik-
keld onder de naam Tesla. Een onderzoe-
Figuur 1: Droomscenario: GPU versus CPU
Optimale inzet van een GPU voor beeldbewerking kan bovenstaande curve opleveren (met fors meer
MIPS voor de GPU dan bij de CPU). Buiten die specifieke grafische toepassing is een GPU beperkt.
Het draaien van niet-grafische toepassingen op een GPU zou een ander plaatje laten zien: de perfor-
mance stort in tot onder de CPU-curve.
Maxwell 13.3 April 2010 45
ker heeft zo voor een relatief bescheiden
bedrag een desktop die qua rekenkracht
kan wedijveren met een supercomputer.
GPU in embedded systemen?Voor professionele technische toepassin-
gen is een GPU alleen maar bruikbaar
in zijn natuurlijke habitat: op een gra-
fische kaart, ingebouwd in een pc, met
standaard software. Mochten GPUchips
al los verkrijgbaar zijn, dan nog zijn ze
moeilijk in embedded systemen in te
passen. Het probleem moet passen bij de
architectuur van de chip, de chip moet
passen in het systeem en het business-
model moet kloppen. Zo heeft de GPU
voor zijn inzetbaarheid bij grafische
toepassingen een korte levensloop en is
daardoor slecht verkrijgbaar op de lange
termijn. De professionele embedded we-
reld is vrijwel altijd beter uit met een
FPGA. Die zijn wel lang genoeg lever-
baar. De keus in FPGA’s is enorm: voor
elk denkbaar probleem is wel een speci-
fieke FPGA te vinden. Deze zijn ook nog
eens uiterst flexibel. De ontwikkelaar
kan namelijk zelf de balans kiezen tus-
sen parallel en serieel, tussen data load
en rekenkracht. Soms zijn er niet zoveel
afzonderlijke berekeningen nodig, maar
moet het wel op 10Gb datastromen ge-
beuren, zoals in een hoge resolutiecame-
ra. Die genereert een bulk data die eerst
wordt bewerkt voordat een pc er iets mee
kan. Misschien zou dat met een GPU
kunnen, maar voor bewerkingen in het
embedded domein kiest Technolution
er altijd voor om de datastroom met een
FPGA te bewerken. Juist vanwege voor-
genoemde zaken.
Toekomst GPUEen van de grootste beperkingen van
vandaag in het parallel rekenen is de
doorvoersnelheid van het geheugen. Lar-
rabee is de codenaam van een ontwik-
keling bij Intel, die een hybride van een
X86 CPU en een GPU in een smeedt tot
een GPGPU (General Purpose GPU). In
dit geval wordt gebruik gemaakt van vele
volwaardige CPU-kernen met SIMD-
support en een conventionele cache.
Door in de toekomst een CPU- en een
Larabeechip te integreren komen beide
typen processoren op een plak silicium
te zitten en kunnen de verbindingen erg
kort en dus snel zijn. Dit geeft veel flexi-
biliteit en performanceverbetering voor
een breder scala rekenintensieve toepas-
singen. De performanceclaims van Intel
worden door de bestaande GPU-specia-
listen met de nodige scepsis benaderd.
Als deze combinatiechip slaagt, zou hij
het einde van de losse grafische kaart
kunnen inluiden en dus geheel nieuwe
toepassingen voor de GPU in beeldbren-
gen.
Dit artikel is overgenomen uit Objective
12, het magazine over innovatie en tech-
nologie van Technolution op management-
niveau.
Hier invoegen: pagina45Technolution_half.pdf
46 Maxwell 13.3 April 2010
Buitengrenzen van TU DelftNa mijn bijna 46-jarige carriere (van student tot hoogleraar-directeur onderzoeksinsti-
tuut IRCTR) op TUDelft wil ik u deelgenoot maken van mijn persoonlijke ervaringen
met TUDelft buitengrenzen.
Wat merk je van TUDelft buitengrenzen als je bezig bent met het afstuderen. Eigenlijk
niet zoveel. Je volgt nog enkele colleges, je hebt contacten met je medestudenten, men-
tor en afstudeerhoogleraar. Dan ga je een baan zoeken. Al snel kom je erachter, dat er
TUDelft buitengrenzen zijn. Bij bedrijven waar je solliciteert zitten veelal oud TUDelft
studenten (alumni) op belangrijke posities. Alumni bepalen zo de beeldvorming van de
TUDelft buitengrens voor de pas afgestudeerde.
Contacten met onderzoeksorganisaties lopen veelal via daar werkende onderzoekers
met een ir. titel. Dan blijkt, dat alumni gaarne bereid zijn jou te woord te staan. Tevens
wordt snel duidelijk dat meerdere alumni een behoorlijke carriere maken en dat zij met
trots de relatie duiden tussen hun positie en de TUDelft opleiding.
ColumnAuteur: Prof. Dr. Leo Ligthart
Mijn passie is internationaal onderzoek. Al in het begin van de 70-tiger jaren had ik interesse voor in-
ternationaal contractonderzoek; laten we zeggen 3e geldstroom onderzoek; alhoewel, termen als 1ste,
2de en 3de geldstroom bestonden toen nog niet. Mijn bevoegdheid tijdens projectonderhandelingen
bestond eruit dat indien er problemen waren, ik mijn ‘professor-baas’ moest bellen. Kortom, er waren
weinig beperkingen.
Als hoogleraar heb ik mijn buitengrenzen verder kunnen verleggen wetende dat kennis over een
bepaald universitair vakgebied geen grenzen kent. Dankzij samenwerking met geselecteerde sterke
internationale onderzoeksteams zijn mogelijke oplossingen van grote maatschappelijke problemen
haalbaar. Deze problemen zijn meestal multi-disciplinair en vereisen een leerstoel (ja zelfs een TU-
Delft) grens-overschrijdende aanpak en de opzet van zogenoemde internationale doorbraakprojecten.
Om mijn multi-disciplinair onderzoek maximaal te faciliteren is het onderzoeksinstituut IRCTR (In-
ternational Research Centre for Telecommunications and Radar) opgericht. IRCTR kreeg erkenning
van het Ministerie van OC&W en werd vereerd met een bezoek van de toenmalige minister Ritzen.
Inmiddels zijn er internationale IRCTR branches en vele samenwerkingsverbanden met bedrijven en
universiteiten. Hierdoor kunnen studenten en promovendi, niet alleen uit Delft maar ook van partner
universiteiten, participeren in internationale programma’s.
Wat is mijn buitengrens: niet de universiteit te Delft of de campus van Delft, maar de partnerinstitu-
ten. Als ik nu een werkbezoek breng binnen of buiten Europa zijn er altijd oud-afstudeerders, die in
eerdere projecten hebben meegedaan en die het op prijs stellen om blijvende contacten met TUDelft
en IRCTR in het bijzonder te onderhouden. Mijn TUDelft buitengrenzen liggen bij al mijn oud-
studenten, -promovendi en -onderzoekers, die als ambassadeurs van TUDelft op weg zijn naar de
top van wetenschappelijke instituten danwel bedrijven of deze positie reeds bereikt hebben. Zij allen
waren eens student en hebben bijgedragen aan de over de grens erkende internationale status van het
TUDelft radar- en telecommunicatie onderzoek.
Maxwell 13.2 January 2010 47
HIER INVOEGEN:PAGINA47ASML.PDF
48 Maxwell 13.2 January 2010
HIER INVOEGEN:PAGINA48FARNELL.PDF